Publication number | US8038616 B2 |
Publication type | Grant |
Application number | US 11/204,492 |
Publication date | Oct 18, 2011 |
Filing date | Aug 16, 2005 |
Priority date | May 30, 2003 |
Fee status | Paid |
Also published as | US20060052699 |
Publication number | 11204492, 204492, US 8038616 B2, US 8038616B2, US-B2-8038616, US8038616 B2, US8038616B2 |
Inventors | Bjørn A. J. Angelsen, Rune Hansen, Øyvind Krovel-Velle Standal |
Original Assignee | Surf Technology As |
Export Citation | BiBTeX, EndNote, RefMan |
Patent Citations (40), Non-Patent Citations (6), Referenced by (17), Classifications (36), Legal Events (2) | |
External Links: USPTO, USPTO Assignment, Espacenet | |
This application is a continuation-in-part of application Ser. No. 11/189,350, filed Jul. 25, 2005, which is a continuation-in-part of application Ser. No. 10/864,992, filed Jun. 10, 2004, which is a continuation of U.S. patent application Ser. No. 10/851,820, filed May 21, 2004, which claimed priority from U.S. Provisional Patent Application Ser. No. 60/475,222, filed May 30, 2003.
1. Field of the Invention
This invention relates to methods and systems for imaging of spatial variation of acoustic parameters of an object and particularly gas bubbles and high density scatterers in the object. The methods have applications in a variety of fields with a variety of objects, for example ultrasound imaging of biological tissues and fluids, acoustic imaging of geologic structures, and detection of objects in water with SONAR.
2. Description of the Related Art
Acoustic imaging is used in a variety of applications, such as medical ultrasound imaging of internal organs, SONAR imaging of fish, sea animals and other objects in the sea, imaging of geologic structures for various purposes such as studies of archeological digs and surveillance of oil wells. A wide range of frequencies of the transmitted acoustic pulse are used for different applications, ranging from infrasound for imaging of some geological structures to ˜100 MHz ultrasound imaging of some biological and microscopic structures. Despite this spread of applications and imaging frequencies the imaging methods are very similar for all applications. In this patent we therefore refer to these imaging methods in general as acoustic imaging, whereas a large part of the applications, especially medical applications, will use inaudible ultrasound frequencies from ˜20 kHz to well into the ˜100 MHz range. Where the imaging frequencies are in the ultrasound range, such as medical imaging, we also will use the term ultrasound imaging, not limiting the methods to ultrasound frequencies and medical applications only.
Despite the widespread use of acoustic imaging, current images are noisy, require large skills for the interpretation, and provide limited quantitative information about the objects. This provides problems for differentiation of object structures and estimating quantitative object properties. With SONAR imaging in water it is for example often difficult to differentiate objects close to or on the sea bed, like fish or other sea animals or mines, from the seabed echoes. Similarly, in geologic imaging it can be difficult to determine material compositions of the geologic structures. In medical applications it can be difficult to differentiate structures like a tumor or atherosclerotic tissue from normal tissue. Important reasons for this are described below.
Spatial variations in the linear acoustic properties of the object (mass density and compressibility) are the physical basis for acoustic imaging. However, with large variations of the acoustic properties in complex structures, the following effects will degrade the images:
i) Interfaces between materials with large differences in acoustic properties can give so strong reflections of the acoustic pulse that multiple reflections get large amplitudes. Such multiple reflections are termed pulse reverberations, and add a tail to the propagating acoustic pulse, which shows as noise in the acoustic image.
ii) Variations of the acoustic velocity within the complex object structures produce forward propagation aberrations of the acoustic wave-front, destroying the focusing of the beam main lobe and increasing the beam sidelobes.
The reduced focusing of the beam main lobe by the wave-front aberrations reduces the spatial resolution in the acoustic imaging system. The pulse reverberations and the increase in beam side lobes by the wave-front aberrations, introduce additive noise in the image, which reduces the ratio of the strongest to the weakest scatterer that can be detected in the neighborhood of each other, defined as the contrast resolution in the image. This noise is termed acoustic noise as it is produced by the transmitted acoustic pulse itself. Increasing the transmitted pulse power will hence not improve the power ratio of the signal to the noise of this type, contrary to what is found with electronic receiver noise.
In echocardiography for example, pulse reverberation noise can obscure images of the apical region of the heart, making it difficult to detect apical thrombi, and reduced contraction of the apical myocardium. Further by example, in carotid imaging reverberation noise can obscure detection and delineation of a carotid plaque. Similar to these examples, the pulse reverberation noise limits the detection of weak targets and differentiation of small differences in image contrast in all aspects of acoustic imaging.
2^{nd }harmonic imaging is a method to reduce the image degrading effect of the pulse reverberations in structures close to the acoustic source like the human body wall, because the 2^{nd }harmonic content in the pulse accumulates as a function of depth and is hence very low as the pulse passes the near source structures like the body wall. However, the sensitivity with 2^{nd }harmonic imaging is less (˜−20 dB) than with 1^{st }harmonic imaging, which limits maximal image depth, particularly in dense objects like geologic structures and biological objects like the liver, kidneys, breast, etc, and for blood velocity imaging. For real time 3D imaging one wants a broad transmit beam that is covered with many parallel receive beams to increase volume image rate. Such broad 2^{nd }harmonic transmit beams are difficult to obtain due to reduced 1^{st }harmonic amplitude in broad beams, which produces problems for 2^{nd }harmonic imaging with multiple parallel receive beams used in real time 3D imaging. This is especially true for sparse acoustic arrays where the number of elements that generates the transmit beam is limited.
In medical applications, tissue diseases like tumors and atherosclerosis of an artery wall, affect the acoustic parameters of the tissue, such as the shear modulus, the bulk compressibility, and the acoustic absorption. The variations of these properties are mainly produced by in-growth of foam cells, fat, or connective tissue fiber molecules, but also through segregation of calcium in the tissue. The in-growth of connective tissue increases the acoustic absorption and the shear modulus, the latter producing an increased stiffness to palpation that can be observed by touching the tissue. Much work has been done on estimation of the shear modulus by using ultrasound bulk waves to register the displacement of shear waves in the tissue in methods often referred to as elastography, also referred to as remote ultrasound palpation. However, to date these methods have found limited clinical application, and there is still a great need for improved differentiation of such tissue changes with ultrasound.
In breast tumors, segregated micro-calcifications are today detected with X-Ray mammography, as an indication of a malignant tumor. These micro-calcifications are so small that the scattered ultrasound signal from them is buried in the signal from surrounding tissue, and they are not detected with current ultrasound imaging. Hence, it is a need to improve ultrasound imaging to also detect such micro-calcifications. Micro-calcifications in atherosclerotic plaque also give information about the stability of the plaque and improved imaging of these micro-calcifications are needed.
Several diseases also affect the blood perfusion through the tissue, for example through angiogenesis or necrosis of the micro-vasculature in malignant tumors, or reduced blood flow due to vascular stenosis or thrombosis both in the coronary arteries of the heart and in peripheral vessels. The blood velocities in the micro-vasculature and small vessels are so small that they cannot be detected with ordinary, non-invasive ultrasound Doppler techniques. Ultrasound contrast agents in the form of solutions of small micro-bubbles (diam ˜3 μm) have therefore been developed to improve ultrasound imaging of the micro-vasculature and also to estimate the blood perfusion through the tissue. The micro-bubbles are injected into the blood stream and provide highly increased and nonlinear scattering of the ultrasound from the blood. They hence highly increase the nonlinear scattering from the tissue that contains such micro-bubbles, where in special cases single micro-bubbles can be visualized in dense tissues and provides a potential for molecular ultrasound imaging with tissue specific targeted contrast bubbles. Such micro-bubbles can also provide useful image enhancement when injected into other body fluids, for example the insterstitial fluid to trace lymphatic drainage to sentinel lymph nodes, or in the urinary system for targeted attachment of bubbles to tumor tissue, or other. During decompression in diving and space activities, micro gas bubbles often form spontaneously in the tissue causing decompression sickness, and it is a need for early detection of such gas bubbles to improve decompression profiles and avoid decompression sickness in personnel under such operations, and even to monitor formation of such bubbles as an early warning during activity.
During production in an oil well, one for example wants to monitor changes in the geological structures around the oil wells, for example to monitor the amounts of oil or gas in the sand stone, observe the boundary between gas/oil and water, and observe any structural slides in the neighborhood of the well. The acoustic properties of the structures, and particularly the nonlinear component of the acoustic properties, are influenced by the amount of gas, oil, or water in the porous rock. Acoustic imaging of the structures surrounding the oil well can be done from acoustic transducers in the oil well. Utilizing imaging methods that provide quantitative acoustic data from the object hence allows detections of the amount of gas, oil, or water in the structures surrounding the oil well.
In detection of fish or sea animals or other objects close to the seabed it is often difficult to differentiate between the echoes from the seabed and the object, particularly with side looking beams. The swimming fish or sea animals has a gas filled bladder or lungs that has quite different and nonlinear acoustic properties compared to those of the seabed. These differences in acoustic properties can with methods according to the current invention be used to differentiate overlapping echoes from such objects and the seabed. The methods can also be used to enhance small solid structures, like a mine, on a softer seabed or in soil, similar to detection of micro-calcifications in a tumor.
There is hence a great need for improved acoustic imaging that reduces the image noise, and enhances the image contrast for variations in object properties,
Methods according to the current invention reduce the image noise, and enhance the image contrast for variations in object properties by transmitting dual frequency band acoustic pulse complexes composed of overlapping high and low frequency pulses into the object. Dual frequency band ultrasound pulses have previously been used in medical ultrasound imaging for various purposes, where in M-mode and Doppler [Br Heart J. 1984 January; 51(1):61-9] simultaneous transmission was used of a 3 MHz pulse and a 1.5 MHz pulse with fixed phase relation between the pulses, for optimal M-mode imaging of the heart (3 MHz pulse) and Doppler blood velocity measurements (1.5 MHz pulse) to interrogate cardiac defects. A concentric annular transducer arrangement was used, where the 3 MHz M-mode ultrasound pulse was transmitted and received by the central transducer disc, while the 1.5 MHz Doppler ultrasound pulse was transmitted and received by a surrounding annular element.
The use of dual band transmitted pulses is also described in U.S. Pat. No. 5,410,516, for improved detection of ultrasound contrast agent micro-bubbles. In this patent, simultaneous transmission of two ultrasound pulses with different center frequencies is described, where the scattered pulses from the micro-bubbles contain sums and differences of the transmitted frequencies produced by the nonlinear scattering from the micro-bubbles, and these sum and difference frequencies are used for the detection of the micro-bubbles.
A similar use of dual band pulses is described in U.S. Pat. No. 6,312,383 for detection of ultrasound contrast agent, where the phase between the two bands is changed between transmissions. This can be viewed as a special case of U.S. Pat. No. 5,410,516, where the change in phase of the low frequency pulse can be viewed as a beat between the low frequency and the pulse repetition frequency.
However, although both the last two patents use nonlinear scattering with dual band pulses for detection of contrast agent in tissue, the presented patents both fail to recognize the nonlinear effect of the low band pulse on the forward propagation velocity of the high band pulse, which in the practical situation will limit the suppression of the tissue signal in relation to the contrast agent signal. The patents also do not recognize how the nonlinear scattering from ordinary tissues or other objects can be retrieved. Accumulative nonlinear forward propagation effects will produce similar signal characteristics for the strong, linear scattering from the tissue, as for the local, nonlinear scattering from micro-bubbles and tissues. This effect will mask the local, nonlinear scattering from micro-bubbles and other object parts and limit the contrast to tissue signal power ratio (CTR). Presence of gas and micro-bubbles in a region also heavily increases the forward, accumulative, nonlinear propagation effect and makes the linear scattering from the object beyond such a region highly mask the scattering from gas and micro-bubbles in the object. This phenomenon for example highly affects imaging of contrast agent in myocardium with pulses that passes the ventricle with contrast agent before entering the myocardium, and can for example falsely indicate perfusion in an ischemic myocardium. It will also affect the differentiation between gas and oil past a region with gas in geologic structures.
The current invention differs from the prior art in that it utilizes the nonlinear effect on the propagation velocity for the high frequency pulse by the low frequency pulse, and an understanding of this effect, in the formation of image signals based on the high frequency propagated and scattered signals. This allows a separation of the accumulative nonlinear effect on the signals from the effect of the local nonlinear object parameters, hence allowing estimation of local nonlinear object parameters, which is not possible by prior art. The invention further devices a method for separation of the accumulative effect of acoustic absorption on the signals, enabling the estimation of the local acoustic absorption parameters of the object.
The methods have applications to acoustic imaging both with back-scatter signals, and computerized reconstruction imaging based on angular scattering and/or forward transmission measurements.
Dual band acoustic pulse complexes with pulse components both in a low and a high frequency band that overlaps in the time domain, are transmitted towards the region of the object to be imaged. The nonlinear manipulation of the object scattering and propagation properties for the high frequency pulse by the low frequency pulse is utilized in the process of forming image signals. The high frequency components of the received signals are processed to give the image parameters/signals, and the low frequency components in the received signals can be removed through filtering, for example directly in the receive transducer array.
The processing according to the invention is part of the complete processing necessary to form images, where additional processing that is necessary to form the final image but not disclosed in this invention, is part of the open knowledge. For backscatter imaging, the methods are used to form radial image lines where 2D and 3D images are obtained by lateral beam scanning according to known methods. The radial image lines can be the signal envelope for structural images, Doppler measurements of radial scatterer displacement and displacement velocities, radial displacement strain or displacement strain rates of relative scatterer movements, or fast time (depth time) spectral parameters for object characterization. Parallel transmit and/or receive beams can be used to obtain multiple radial image lines in parallel to speed up the frame rate. With computer tomographic (CT) image reconstruction the methods provide improved measurements for the reconstruction, with reduced pulse reverberation noise and nonlinear image parameters that provide complementary information.
The invention devices several methods for improved imaging with increasing number of pulses required to form an image, with a complementary reduction in image frame rate, but with increasing image quality. The invention therefore further devices an instrument for operation of two or more of the methods and procedures for optimal selection of the methods for best performance of the instrument under given constraints, such as frame rate, image quality, a combination of frame rate and image quality, etc.
In a 1^{st }method according to the invention, the high frequency pulse propagates on a negative spatial gradient of the low frequency pulse oscillation, so that the back of the high frequency pulse gets a higher propagation velocity than the front of the pulse, due to the nonlinear effect on the propagation velocity by the low frequency pulse. This produces a cumulative spatial compression of the high frequency pulse as it propagates into the object, increasing the frequency and the bandwidth (i.e. shortens the length) of the high frequency pulse, in addition to the nonlinear self-distortion of the high frequency pulse producing harmonic components in the pulse. This increase in frequency given by the pulse length reduction, is counteracting the lowering of the pulse center frequency by the frequency dependent absorption in the object, hence providing a higher received center frequency than when this method is not utilized.
As the amplitude of the low frequency pulse is greatly reduced in the first reflection, multiple scattered pulses will not have this same length compression from the nonlinear effect on the propagation velocity for the high frequency pulse by the low frequency pulse, and will due to absorption drop to lower frequencies than first order scattered pulses with the same propagation lag, and can hence be filtered away producing a markedly suppression of the pulse reverberation (multiple scattering) noise, similar to 2^{nd }harmonic imaging but with 1^{st }harmonic sensitivity allowing deeper imaging and the use of higher acoustic imaging frequencies than with 2^{nd }harmonic imaging, improving spatial resolution. It is also simpler to obtain broader transmit beams allowing the use of more parallel receive beams, compared to 2^{nd }harmonic imaging, allowing higher image frame rates for 2D and especially 3D imaging. This is especially true when sparse arrays are used for the transmit beam, where it is difficult to obtain high enough amplitudes for adequate harmonic pulse self-distortion due to the limited number of array elements. The nonlinear pulse compression is also interesting when the 2^{nd }harmonic band of the received signal is used for imaging, as the frequency down-sliding of absorption is counteracted, providing higher frequencies and shorter pulses at deep ranges with improved resolution. The invention also devices this type of pulse compression for observation pulses of object displacement from radiation force push pulses, for frequency separation between the observation and the push pulses. The invention also devices placement of the high frequency pulse close to a peak in the low frequency pulse of a transmitted pulse complex, to allow for higher transmitted amplitude of the high frequency pulse with limitations in the Mechanical Index (MI) in the object.
In a 2^{nd }method according to the invention one transmits two or more dual band pulse complexes in sequence for each radial image line, where the high frequency pulse is found close to the peak or trough of the low frequency pulse, and where the frequency and/or phase and/or amplitude of the low frequency pulse vary for each transmission, to nonlinearly manipulate the acoustic scattering and forward propagation properties of the object for the high frequency components. The nonlinear manipulation of the forward propagation velocity is also with this method utilized in the process of forming image signals.
One can for example with this method also form a 1^{st }image signal, Eq. (14), with highly suppressed pulse reverberation noise with 1^{st }harmonic sensitivity, to be utilized with the same advantages as for the single pulse described above. The invention further devices to estimate the nonlinear propagation delays, which provides a 1^{st }quantitative nonlinear image parameter, Eq. (27), which is a quantitative nonlinear forward propagation parameter, as a combination of the differential of the estimated nonlinear propagation delays and an estimate of the amplitude of the low frequency pulse. The frequency of this pulse can be chosen so low (typically ˜⅕- 1/20 of the high, imaging frequency) that differences in acoustic power absorption between different objects and individuals can be neglected, and the low frequency pulse amplitude can be estimated from simulations or measurements in water or oil mixtures. The invention further devices a method of estimation of the local absorption coefficient through a combination of the estimated nonlinear propagation delays according to the 2^{nd }method (and also according to the 3^{rd }and 4^{th }method described below), and the radial gradient of the nonlinear propagation delay, and the center frequency in the high frequency received signal under the 1^{st }method, and the radial gradient of said center frequency.
The reduced reverberation noise in the received signals according to the invention greatly helps the estimation of corrections for wave front aberrations, for example as described in U.S. Pat. No. 6,485,423, U.S. Pat. No. 6,905,465 and U.S. patent application Ser. No. 10/894,387, in conjunction with the current invention. The invention also gives an approximate estimate of delay corrections for the wave front aberrations, derived from the nonlinear propagation delays estimated for the signals from each element or sub-aperture signal further defined in the specification below.
The invention further devices to correct the received high frequency signals with the nonlinear propagation delay estimates in the process of forming image signals. One is then able to highly suppress the linearly scattered signal from the object in the process and provide a 2^{nd }image signal, Eq. (19, 28) which is the nonlinearly scattered signal that shows local, nonlinear properties of the object on a scale less than the high frequency wave length, whereas the nonlinear propagation parameters show nonlinear object properties on a scale larger than a couple of high frequency wave lengths. The nonlinear signal then provides image contrast to rapid changes in object structures with improved differentiation of the structures. The nonlinear scattering is specially high at interfaces between materials with large differences in compliance, such as at interfaces between soft biological tissue and stiffer tissue like connective or muscular tissue or solid materials like calcifications, or between low and high compliance object parts like fat or micro gas bubbles in biological tissues, gas in porous rock, and fish swim-bladder or sea animal lungs and surrounding water and tissue, hence improving the characterization of the object structures.
Gas bubbles are often found naturally in the object as for example a fish swim bladder or sea animal lung, gas bubbles in porous rock, micro bubbles formed spontaneously during decompression in tissue, or micro bubbles injected into the object as a contrast agent. The bubble compression dynamics with acoustic pressure waves is described by a differential equation, providing a resonant acoustic scattering with a frequency dependent phase lag between the incident and the scattered wave, contrary to scattering from solids or fluids where the frequency variation of this phase is practically negligible. The resonance frequency and hence this phase lag of the scattered signal for the high frequency pulse from a gas-bubble is also manipulated by the low frequency pulse, in addition to the amplitude of the signal, which allows extraction of most of the scattered power from the gas-bubbles with this 2^{nd }method (not only the harmonic components), and significantly increases the CNR (Contrast to Noise Ratio) relative to existing methods. With good estimation of the nonlinear propagation delay corrections and also amplitude corrections, the methods according to the invention will strongly suppress the linearly scattered signal from the object, and significantly increase the COR (Contrast to Object Ratio) relative to existing methods. Contrary to state-of-the-art contrast agent detection methods, like harmonic imaging, pulse inversion, or power Doppler, the methods according to the current invention can use higher acoustic frequencies relative to the resonance frequency of the bubble, with improved spatial resolution. One can also use lower pulse amplitudes (lower Mechanical Index (MI)), which avoids destruction of medical contrast agent bubbles. This is important for imaging of tissue-specific targeted micro-bubbles where a limited group of micro-bubbles adheres to selected tissues, for example tumor tissues, atherosclerotic plaque, thrombi, etc. where it is important to image the bubbles without destruction.
A cloud of micro-bubbles in biological tissue, fluids, or porous rock will have strong, nonlinear effect on the propagation velocity of a through-passing pulse, and in such cases it is especially important to provide corrections for the nonlinear propagation delays for good suppression of the linearly scattered object signal beyond the cloud of bubbles. With this delay correction, the invention provides a separation between the accumulated nonlinear forward propagation delay, and the local, nonlinear scattering, contrary to what is found with other methods like harmonic or pulse inversion imaging, and provides a great advantage for suppression of object image signal when imaging gas-bubbles past the cloud, for example in the distal myocardium in medical imaging, detection of gas past a gas region in geological imaging, or detection of fish or sea animals past a school of fish or sea animals. If no or limited corrections for the nonlinear propagation effects are done in these cases, the linearly scattered signal from objects in regions beyond gas bubbles will show similar properties as the scattering from the gas bubbles, hence masking the detection of gas bubbles in these regions. This can for example falsely indicate blood perfusion in an ischemic region of myocardium, gas in geological structures, and fish or other sea animals in the water.
Aside from the multiple medical use of imaging of ultrasound contrast agent micro-bubbles, imaging of micro-bubbles according to this method in decompression situations found in diving and space activities, can be used to monitor formation of such bubbles to study and develop decompression profiles, or as an early safety alarm against sickness during decompression.
In further processing according to this 2^{nd }method, the delay corrected high frequency signals are combined along the pulse number coordinate to provide a 3^{rd }image signal, Eq. (17,29), the linearly scattered signal. This linearly scattered signal has the same attenuation due to power absorption as the 2^{nd }image signal, the nonlinearly scattered signal. Through a combination of the nonlinearly and the linearly scattered signals and the estimate of the low frequency pulse amplitude presented above, the invention presents a 2^{nd }quantitative nonlinear image parameter, Eq. (30), which is a quantitative nonlinear scattering parameter. This 2^{nd }quantitative nonlinear parameter then represents the spatial fluctuations in the nonlinear object parameters on a scale˜smaller than the high band wave length, while the 1^{st }quantitative nonlinear parameter, Eq. (27), reveal a spatial average of the nonlinear object parameters on a scale˜larger than the high band wave length. The backscatter and the forward propagation hence reveal two different quantitative image parameters that can be visualized for increased information about the object characteristics. The quantitative nonlinear parameters hence improve differentiation of object structures, and also open for object characterization with the method, which is useful in diagnosing tumors and atherosclerotic plaque in medical applications, assessing gas, oil or water in geologic structures, and assessing amount and size of fish and sea animals in water. Calibration of the thermal variation of these quantitative parameters also opens for local temperature estimation with ultrasound, for example to be used for guidance of hyper- or hypo-thermal treatment of tumors in medicine. It further provides new methods of quantifying contrast agent volume in tissue, blood perfusion through the tissue, and relative volume of gas and oil in geological structures.
With non-moving, temporary stationary objects, one can for example transmit two pulses with different frequency and/or phase and/or amplitude of the low frequency components, and combine the scattered or transmitted signals from these pulses to suppress the pulse reverberations and estimate the nonlinear object parameters. When the object and acoustic probe move relative to each other, it is advantageous to transmit more than two pulses for each radial image line to adequately suppress the linearly scattered signal or suppress the pulse reverberation noise with multiple pulses. For example, one can transmit a set of K pulses, all with the same phase of the high frequency components, but with different frequencies and/or phases and/or amplitudes of the low frequency components for each pulse. The back-scattered signals from these pulses are combined in a pulse-to-pulse high pass filter that suppresses the pulse reverberations and lets through the 1^{st }order scattered signal components. With estimations and corrections for the nonlinear propagation delays, and optionally also amplitude corrections, before the high pass filter, one can extract the local, nonlinearly scattered signal from the object or the scattered signal from gas bubbles, and quantitative nonlinear propagation and scattering parameters of the object.
In this 2^{nd }method the pulse reverberation noise (and to a small degree the nonlinear signal components themselves), introduce errors in the estimates of the nonlinear propagation delays. These errors limit the suppression of the linearly scattered signal when estimating the nonlinearly scattered signal. To efficiently remove the effect of the pulse reverberation noise on the estimation of the nonlinear delay corrections, one can use the 2^{nd }harmonic component of the scattered signals with the 2^{nd }method according to the invention, or one can according to a 3^{rd }method of the invention transmit at least three pulses with different frequencies and/or phases and/or amplitudes of the low frequency pulse, as described in relation to Eqs. (42-44). The 3^{rd }method still has influence of the nonlinear scattering in the estimation of the nonlinear propagation delays. In a 4^{th }method according to the invention described in relation to Eqs. (45-48) one transmits at least 4 pulses with 4 different levels of frequencies and/or phases and/or amplitudes of the low frequency pulse, enabling the estimation of the nonlinear propagation delays, the linearly scattered signal, and the nonlinearly scattered signal, with minimal interference between each other and from reverberation noise.
It is also possible to use the 2^{nd }harmonic band (or any harmonic band) of received signal both for the image formation according to any of the 2^{nd}, 3^{rd}, and 4^{th }methods, instead of the 1^{st }harmonic band of the received signal as described above. The advantage is added suppression of the pulse reverberation noise both in the image signals, and in the estimation of the nonlinear propagation delay as described above. The disadvantage is less sensitivity at deeper ranges reducing the image depth for the same image frequency.
With electronic steering of the beam direction one would typically use the same beam direction and transmit focus for all the transmit pulses for each radial image line and depth range, where the received signals are combined to suppress the linearly scattered object signal for that image line. Typical filtering schemes that are used are FIR-type filters or filters with time variable impulse response like orthogonal decomposition using for example Legendre polynomials, with filtering along the pulse number coordinate for each depth.
With mechanical scanning of the beam direction, as with annular arrays or 3D imaging, one would typically transmit pulses with variations in the frequency and/or phase and/or amplitude of the low frequency pulse as the beam direction is swept continuously, feeding the signal for each depth to a high pass filter along the pulse number coordinate. The outputs of the high pass filters are then sampled for each depth and radial image line to estimate the signals and image parameters to be used for image reconstruction along that radial image line in the depth range.
The invention further presents basic designs of imaging instruments that operate according to the methods according to the invention. As the number of pulses per radial image line, together with image quality and information, increases with the order of the methods, the frame rate decreases with the order of the methods. In a most advanced version, the instrument can operate more than one of the methods with procedures for optimal selection of the methods for best performance of the imaging under given constraints. Typical constraints are a minimal frame rate, minimal requirements on image quality etc.
As a last point, the invention provides a design procedure of transducer arrays that minimize the nonlinear effect on the propagation delay of the high frequency pulse by the low frequency pulse. With low amplitudes (˜50 kPa) of the low frequency pulse components, such transducer arrays can allow imaging of medical ultrasound contrast agents or decompression micro bubbles with a limited but still interesting suppression of the linearly scattered signal from the object, without correcting for the nonlinear propagation delays of the high frequency pulse produced by the low frequency pulse.
Other objects and features of the present invention will become apparent from the following detailed description considered in conjunction with the accompanying drawings. It is to be understood, however, that the drawings are designed solely for purposes of illustration and not as a definition of the limits of the invention, for which reference should be made to the appended claims. It should be further understood that the drawings are not necessarily drawn to scale and that, unless otherwise indicated, they are merely intended to conceptually illustrate the structures and procedures described herein.
In the drawings:
Acoustic bulk waves in homogeneous materials are in the linear regime governed by a linear wave equation where the bulk wave propagation velocity c_{0 }is determined by the mass density ρ_{0 }and the bulk compressibility κ_{0 }of the homogeneous propagation medium. The bulk compressibility is in the linear approximation of bulk elasticity defined through the relative volume compression of the material as
where δV is the relative volume compression of a small volume ΔV subject to the pressure p, and ψ is the particle displacement in the material so that −∇ψ is the relative volume compression.
The scattering of acoustic bulk waves from objects is produced by spatial fluctuations in the compressibility and mass density of the object. We denote the spatially varying mass density and compressibility for low pressure amplitudes as ρ_{0}(r) and κ_{0}(r), where r is the spatial coordinate. The linear back-scattering coefficient from a local point r is then
where ρ_{0a}(r) and κ_{0a}(r) are spatial averages of the mass density and bulk compressibility on a scale˜a couple of wave lengths λ of the acoustic pulse, and e _{i }is the unit vector in direction of the incoming wave, and e _{s }is unit vector in the direction one observes the scattered wave, as illustrated in
The spatial variation in the average mass density and bulk compressibility, ρ_{0a}(r) and κ_{0a}(r), produces a spatial variation in the acoustic propagation velocity as
This spatial variation of the propagation velocity is responsible for aberrations of the wave front, in biological imaging specially found in the body wall, but also throughout the whole of some objects, like the breast and glands that contain regions of fat or connective tissue. In geologic imaging large variations in the propagations velocity is often found throughout the whole object.
Typical values for soft tissues and fluids are κ_{0}˜400·10^{−12 }Pa^{−1 }with a typical acoustic pulse amplitude of p˜10^{6 }Pa, which gives δV/ΔV˜0.4·10^{−3}. A volume compression produces an increase in the mass density as δρ/ρ_{0}=κ_{0}p˜0.4·10^{−3}. Rock shows lower compressibility, while water and oils shows similar compressibilities as tissues. Similarly, when the object is compressed, there is a decrease in the bulk compressibility which together with acoustic absorption in the object modifies Eq. (1) as [1]
where β_{n}=(1+B/2A)˜5 is a nonlinearity parameter related to the commonly defined parameters B and A for the nonlinear bulk modulus [1]. The temporal convolution between the pressure waveform and h represents the frequency dependent acoustic power absorption in the material. The first term describes a nonlinear bulk compressibility influenced by the pressure where a differentiation of this term a reference pressure p_{0 }gives
which gives a relative variation of the compressibility with pressure as δκ_{n}/κ_{0}=−2β_{n}κ_{0}p_{0}. The nonlinear variation in bulk compressibility is hence 2β_{n}˜10 times higher than the nonlinear variation in mass density produced by the pressure, where for p˜1 MPa we get δκ_{n}/κ_{0}=−2β_{n}κ_{0}p_{0}˜4·10^{−3}.
The nonlinear variation of the mass density and the compressibility produces a nonlinear modification of both the scattering and the forward propagation velocity of the wave, and the invention utilizes these effects to reduce pulse reverberation noise, increase image contrast for various object structures, micro-calcifications, and gas bubbles, and produce quantitative acoustic image parameters of the object, micro calcifications, gas bubbles and gas filled regions. In the following we describe example embodiments of the invention with reference to the Figures.
In a 1^{st }method according to the invention, we utilize time compression and expansion of the high frequency pulse by the low frequency pulse to manipulate the center frequency and the bandwidth of the forward propagating pulse within the object. This time compression is produced by the pressure dependency of the forward propagation velocity, which can be approximated as
where β_{na }and κ_{0a }are local, spatial average values over a couple of wavelengths at zero pressure as defined above. To further illustrate this principle, we refer to
The pressure dependent propagation velocity produces an accumulatively increasing forward propagation distortion of the pulse determined by the actual pulse pressure, which is the sum of the low and the high frequency pulse pressures, which after a propagation distance r produces the distorted low frequency pulse 103 with the distorted high frequency pulse 104. The distortion of the high frequency pulse can be separated into a pulse length compression of the zero points of the high frequency pulse produced by the local low frequency pulse pressure, and a pulse shape self distortion produced by the instantaneous high frequency pressure itself. 105 in the
This nonlinear forward propagation distortion of the pulse, is the same effect that produces harmonic components in the forward propagating pulse, that is linearly back scattered from the object, and is used in harmonic imaging of objects, further discussed in relation to Eqs. (10-14) and
To clarify the essential concepts, we first analyze a situation with no beam divergence (i.e. plane waves) and no ultrasound absorption. The pressure difference of the low frequency pulse across a wavelength λ_{1 }of the high frequency pulse is Δp_{0}=λ_{1}∂p_{0}/∂r. This gradient produces a difference in propagation velocity over a high frequency wavelength of Δc_{a}=−β_{na}κ_{0a}c_{0a}Δp_{0 }which produces a compression of the wave with propagation distance r. We note that the same Δc_{a }produces an increase with r in the low frequency pressure gradient proportional to the reduction in the high frequency wavelength, so that the pressure drop along the compressed high frequency wavelength is maintained as Δp_{0 }as illustrated in
The compression/expansion is accompanied by a proportional change in the pulse bandwidth (˜inverse pulse length). According to these formulas, the frequency becomes infinite for z_{sh}=1/β_{na}κ_{0a}p_{00}k_{00 }which is a phenomenon called acoustic shock, where the negative gradient of the low frequency oscillation becomes infinite. In the practical situation acoustic power absorption prevents this situation.
The acoustic absorption also produces a down sliding of the high frequency pulse center frequency while preserving the pulse bandwidth. The compression increase in frequency is only found for the outgoing pulse where the amplitude of the low frequency pulse is sufficiently large, while the absorption down sliding is found both for the outgoing and scattered pulse, which for back scattering gives a propagation distance of 2r. For a Gaussian pulse envelope the down sliding in frequency is given as
Δf _{1a}=−0.36αB _{1} ^{2} r (8)
where B_{1 }is the 6 dB bandwidth of the high frequency acoustic pulse, and α=dBatt/8.686 (mmMHz)^{−1 }is the average, local frequency constant of the absorption coefficient with linear frequency dependency of the absorption. With dBatt=0.05 dB/mmMHz we get α=5.76·10^{−3 }(mmMHz)^{−1}. We should note that with transmission computed tomography imaging, which is discussed in relation to
The typical imaging range, R, for backscatter imaging is limited by the acoustic absorption that increases linearly with frequency. The range R is therefore related to the high frequency acoustic wavelength λ_{1}=c_{0a}/f_{1 }as R˜200λ_{1}-300λ_{1}. For p_{0}=1 MPa, β_{n}=5, κ_{0a}=400·10^{−12 }Pa^{−1}, R=250λ_{1 }and λ_{10}/λ_{00}=10, we get Δf_{1p}=0.458f_{10}, which for f_{10}=10 MHz gives Δf_{1p}=4.58 MHz. The acoustic shock is found for z_{sh}/λ_{10}=796 which is 2.65 times the maximal image range. For B_{10}=5 MHz we get an absorption down sliding of the high center frequency according to Eq. (8) of Δf_{1a}=4 MHz, which balances the compression up-conversion of the frequency, so that the received frequency for the first order back scatter in the absorbing medium is approximately f_{10}=10 MHz for all depths, i.e. as transmitted.
As the pulse compression is produced by the low frequency pulse where as described above the power absorption can be neglected for actual imaging ranges, the pulse length compression with the corresponding increase in the bandwidth is practically independent of the absorption over actual imaging ranges. The absorption down sliding of the center frequency of the high frequency pulse, however, is produced by the absorption of the high frequency pulse. The frequency down sliding is proportional to the square of the absolute bandwidth B_{1 }of the pulse, and the absorption down-sliding preserves the bandwidth. Hence, the combined effect of the nonlinear pulse compression and the absorption down conversion in the above example is a pulse with approximately constant center frequency which maintains the lateral resolution (beam width), but with bandwidth (˜inverse pulse length) that increases with depth which improves the range resolution with depth. Note from Eq. (8) that the increase in B_{1 }width depth produces accelerating absorption down-sliding with depth. We return to more detailed analysis of this situation in relation to Eqs. (35,36).
The forward propagation up-conversion of the high band pulse frequency can be used to improve image resolution at deeper ranges. It can also be used to increase penetration with better resolution at deep ranges where for example one transmits a fairly low frequency that is increased to a higher frequency at deeper ranges by the low frequency pulse, hence reducing the total absorption along pulse path for the obtained high band pulse frequency at the deep ranges. It can also in this aspect be utilized a sliding between the phase of the low and high frequency pulses with propagation distance, with special designs of the low frequency beam profile in relation to the high frequency beam profile as discussed in relation to
The amplitude of the low frequency pulse drops heavily at the first reflection, and the compression/expansion effect on the high frequency pulse by the low frequency pulse is practically negligible after the first reflection as discussed in relation to
A 1^{st }image signal with strong suppression of the pulse reverberation noise, can according to the invention now be obtained with a receiver filter which suppresses lower frequencies with a cut-off frequency that slides with depth range. In
With this method one is hence able to preserve 1^{st }harmonic sensitivity for deep ranges with a suppression of the pulse reverberation noise similar to 2^{nd }harmonic imaging, allowing deeper imaging of dense objects like the liver, the kidneys, the breast, etc with higher frequencies and better resolution. The compression reduction in the high frequency pulse length also improves the range resolution in the image in a way not previously seen. With a beam profile design as discussed above, where the phase between the low and the high frequency pulse slides with depth so that the high frequency pulse is expanded for large depths, it is still possible with proper design to keep the pulse reverberation noise sufficiently separated from the 1^{st }order scattering in the frequency domain, so that the pulse reverberation noise can be sufficiently suppressed with the receiver filter. This modification of the method will hence provide deeper penetration, while maintaining high frequencies for better resolution in the mid to near field. One should however note that in some applications it can be advantageous to use the 2^{nd }harmonic band of the received signal throughout the whole image range, for best suppression of the pulse reverberation noise and also best spatial resolution in the image. The advantage of the pulse compression is then improved range resolution at deep ranges due to the pulse length compression, and also increased 2^{nd }harmonic frequencies at deep ranges producing narrower beams and better lateral resolution. The improved separation between the 2^{nd }harmonic band and the band of the pulse reverberation noise at deep ranges also improves the suppression of the pulse reverberation noise compared to standard 2^{nd }harmonic imaging.
The received high frequency signal is retrieved from the total scattered/transmitted signal (also including the low frequency components) by filtering, for example directly in the receiver transducer. However, the nonlinear propagation distortion of the low frequency pulse component, as illustrated in
By placing the high frequency pulse at the positive peak of the low frequency pulse in the transmitted complex, one reduces the Mechanical Index (MI) of the high frequency pulse (due to lower negative amplitude of the total pulse complex), which allows transmission of higher high frequency amplitudes. This produces higher harmonic distortion of the high frequency pulse, increasing the sensitivity with harmonic imaging with the high frequency pulse, where the receiver filter above is set to select the harmonic bands of the high frequency pulse to suppress pulse reverberation noise.
It should be evident that one for each beam direction can transmit more than one of the pulse complexes in
The method of pulse compression of the high frequency pulse is also useful in situations where the radiation force of ultrasound pulses is used to push the object locally, for example to measure the shear deformation related to the shear modulus of the object, or to improve attachment of targeted contrast agent bubbles to selected tissues. With the current method one can increase frequency separation between the push pulses and the observation pulses, by placing the observation pulses at the negative spatial gradient of the low frequency pulse to increase the receive frequency of the observation pulses. The push pulses can be transmitted as longer pulses with zero low frequency pulse, or a sequence of short high frequency push pulses placed close to a positive spatial gradient of the low frequency pulse for stretching and frequency down conversion of the high frequency push pulses. In both situations the observation pulses gets higher frequencies than the push pulses and can be transmitted shortly after or during a sequence of push pulses, where the separation of the echoes from the observation pulses and the push pulses are done by filtering in the depth (fast) time domain as above.
In a 2^{nd }method according to the invention one transmits two or more pulse complexes with frequency components in a low and a high band which overlap in time, and where the amplitude and/or the phase and/or the frequency of the low frequency pulses vary from pulse to pulse. The method provides another type of received signal with highly suppressed pulse reverberation noise similar to the previous method according to the invention, and also allows imaging of nonlinear scattering parameters in the object, especially imaging of micro calcifications, gas bubbles, and gas filled regions, and also provides quantitative nonlinear scattering and propagation parameters of the object. We start describing the method with reference to
The nonlinear compressibility term is now ˜2β_{n}·2.5˜25 times larger than the nonlinear mass density term and the two terms generally have opposite signs. Imaging with high frequency pulses centered at ω_{1}, will produce a band pass filter in the range coordinate of this parameter in the frequency range around 2k_{1}=2ω_{1}/c as described above.
Further by example, we transmit a 2^{nd }pulse as illustrated in
Due to the nonlinear variation of the propagation velocity with the pressure as given in Eq. (6), the propagation velocity of the high frequency pulse will vary between the pulses in
where s is the distance along the beam axis, t_{0}(r) is the time lag for p_{0}=0, τ(r) is the added nonlinear propagation time lag due to the nonlinear manipulation of the propagation velocity for the high frequency pulse by the low frequency pulse, and p_{0}(s) is the amplitude of the low frequency pulse at the location of the high frequency pulse as a function of depth. We shall in the following refer to t(r) as the nonlinear propagation time lag or nonlinear propagation delay.
The high frequency pulse will also have an accumulative self-distortion as described in relation to
When the phase relations of the high and low frequency components are practically constant along the beam, this time shift will vary monotonously with the local spatial average of β_{na}κ_{a}, as shown in
For an amplitude of the low frequency pulse of p_{0}˜1 MPa we get Δc/c_{0a}=β_{na}κ_{0a}p_{0}˜2·10^{−3}, which for a range R=300λ_{1}=300c_{0a}T_{1 }where T_{1}=1/f_{1 }is the period of the high frequency pulse, we get an added time lag from Eq. (10) of τ(R) up to ˜0.6T_{1}, i.e. close to the period at the high band center frequency. This image range is typical for medical ultrasound imaging, while in water one can have larger ranges, and in geological applications the maximal image range can be below this value. We should note that the difference in nonlinear propagation lag for the positive and negative low frequency pulse in
We now describe how the invention establishes image signals by the nonlinear manipulation of the scattering and propagation parameters of the object for the high frequency pulse, by the low frequency pulse, according to the 2^{nd }method of the invention. We first examine the situation with back scatter imaging and let x_{k}(t) denote the received back scattered signal from transmitted pulse no k, where examples are given in
A mathematical model of the back-scattered 1^{st }harmonic band of the high frequency signal x_{k}(t) where the low frequency pulse switches polarity between transmit pulses, i.e. p_{0k}˜(−1)^{k}, and the scatterers are moving, can be written as
where ω_{d}=−2ω_{1}v_{r}/c_{0 }is the average Doppler shift for scatterers moving with average radial velocity v_{r }away from the transducer in each range cell along the beam. u_{1k}(t) is the complex envelope of the linearly back-scattered scattered signal and u_{nk}(t) is the complex envelope of the nonlinearly back-scattered signal from high frequency pulse no k with positive amplitude p_{0 }of the low frequency pulse. The envelopes vary with the pulse number coordinate k because scatterers and the beam move relative to each other, and scatterers within the range cell can move with different velocities, both producing a frequency broadening of the signal in the slow time coordinate. □(t) is the nonlinear propagation lag as a function of the fast range-time coordinate for positive amplitude of the low frequency pulse. For simplicity, we have used the analytic form {circumflex over (x)}_{k}(t) of the received signal where the physical, radio frequency acoustic signal x_{k}(t)=Re {{circumflex over (x)}_{k}(t)}. The analytic signal can be obtained from the physical signal as {circumflex over (x)}_{k}(t)=x_{k}(t)+iH {x_{k}(t)}=x^{0}/k^{0}(t)exp {iω_{1}t} where H{ } denotes the Hilbert transform of the signal, and {tilde over (x)}_{k}(t) is the complex envelope of the signal.
The 2^{nd }harmonic band can be represented by a similar formula as Eq. (11) where the angular frequency is 2ω_{1}, the Doppler frequency is 2ω_{d}, and the nonlinearly scattered signal is very low and can be neglected except for scattering from micro bubbles. The 2^{nd }harmonic band has suppressed pulse reverberation noise which can help in the estimation of the nonlinear propagation delay which we return to in relation to Eq. (21).
The pulse-to-pulse switching of the nonlinear delay, (−1)^{k}τ(t), has strongest effect in the phase as the phase switching (−1)^{k}ω_{1}τ(t) compared to the delay switching of the envelopes, t−(−1)^{k}τ(t), because the bandwidth of the signal is limited. To visualize the effect of the delay switching on the envelopes we separate the envelopes in an even and an odd function around t, which allows us to express the complex envelope of the received signal as
{tilde over (x)}_{k}(t)={(u _{1k} ^{e}(t,τ(t))−u _{nk} ^{o}(t,τ(t)))+(−1)^{k}(u _{nk} ^{e}(t,τ(t))−u _{nk} ^{o}(t,τ(t)))}e ^{iω} ^{ d } ^{T} ^{ prf } ^{k−iω} ^{ 1 } ^{τ(t)(−1)} ^{ k } (12)
where the superscript e denotes the even components and o denotes the odd components in τ around t. The even component is unchanged by a change in sign of □, while the odd component changes sign. We note that (−1)^{k}=exp {iπk}=exp {ikT_{prf}ω_{prf}/2}, where ω_{prf}=2π/T_{prf }is the angular pulse repetition frequency and hence the angular sampling frequency in the slow time coordinate. Introducing this expression we can further develop Eq. (12) to
For fixed fast time t, the signal is in the slow time coordinate k composed of 4 components (frequency lines) as illustrated in
The lines are generated through a mixing of linear and nonlinear scattering with the pulse to pulse switching of the nonlinear propagation delay, (−1)^{k}τ(t). With no nonlinear delay switching, i.e. τ=0, the odd components become zero and the even components are equal to the original envelopes. The lines 602 and 604 disappear and all the linearly scattered power is contained in the line 601 centered around ω_{d}, while the nonlinear scattered power is contained in the line 603 centered around ω_{d}+ω_{prf}/2.
The effect of the nonlinear propagation delay switching is strongest in the phase of the signal because the limited signal bandwidth, where it produces a frequency mixing with a shift ω_{prf}/2. It also produces a frequency mixing with a shift of ω_{prf}/2 by its participation in the envelopes as t−(−1)^{k}τ(t) through the odd components of the envelopes for the linear scattering, u^{o} _{1k}(t), and the nonlinear scattering, u^{o} _{nk}(t), while the envelope delay switching has no frequency shift effect on the even components of the envelopes. The delay shifting in the phase produces a shifting of part of the linearly scattered power from centered around ω_{d }to the line 602 centered around ω_{d}+ω_{prf}/2 represented by the even component u^{e} _{1k}. A combined switching in the phase and the envelope shifts part of the power from centered around ω_{d }to centered around ω_{d}+ω_{prf}/2 and back to centered around ω_{d }as part of line 604 represented by the odd component u^{o} _{1k}. Similarly, the nonlinear delay switching shifts part of the nonlinear scattered power from line 603 to 604 through the switching in the phase and represented by the even component u^{e} _{nk}, while a combined switching in the phase and the envelope shifts part of the power from centered around (ω_{d}+ω_{prf}/2 to centered around ω_{d }and back to centered around ω_{d}+ω_{prf}/2 as part of line 602 represented by the odd component u^{o} _{nk}.
We note that as τ increases, the amplitude of the line 601 drops ˜cos ω_{1}τ(t), while the amplitude of the line 602 increases ˜sin ω_{1}τ(t) and becomes maximum when ω_{1}τ(t)=π/2, which also gives zero for the line 601. This means that most of the linearly scattered power is moved from being centered around ω_{d }to being centered around ω_{d}+ω_{prf}/2 by the delay switching in the phase, but some linearly scattered power is mixed back to centered around ω_{d }as part of line 604 through the odd component of u_{1k}(t). The same effect is found for the nonlinear scattering, where as τ increases the amplitude of the line 603 drops ˜cos ω_{1}τ(t), while the amplitude of the line 604 increases ˜sin ω_{1}τ(t) and becomes maximum when ω_{1}τ(t)=π/2, which also gives zero for the line 603. Most of the nonlinearly scattered power is moved from centered around ω_{d}+ω_{prf}/2 to centered around ω_{d }by the delay switching in the phase, but some nonlinear power is mixed back to centered around ω_{d}+ω_{prf}/2 as part of line 602 through the odd component of u_{nk}(t).
If the range cell also covers moving blood or other fluids like oil, the linear scattering from moving fluid will produce additional and usually wider spectra where the power is divided between 605 with the original Doppler shifts ω_{d}=−2ω_{1 }v_{r}/c where v_{r }is the spread fluid velocity, and 606 with frequencies ω_{d}+ω_{prf}/2. The mixing of the fluid signal with the switching in the nonlinear propagation delay follows the same rules as for the linear scattering from the object in lines 601 and 602. The nonlinear scattering from the fluid is, however, so week that it will disappear in the noise.
The signal model in Eqs. (11-13) includes only the first order scattered signal, where the out-going high frequency pulse follows the low frequency pulse. With multiple scattering, also called reverberations, of the outgoing high frequency pulse, we get some modifications of the multiply scattered signal from this model. The amplitude of the low frequency scattered pulse is low, and hence its nonlinear delay effect on the propagation velocity of the scattered high frequency pulse, Eq. (6, 10), can be neglected. This is especially important for reverberations of the outgoing pulse in the body wall, where
The reflected original pulse from a deeper reflector 707 is shown as 708 in
We are now in position to describe and discuss how several new image parameters can be extracted from the propagated and scattered acoustic signals according to the 2^{nd }method of the invention, to form new and improved acoustic images of the object structures and velocities of scatterers. The extraction of the parameters can be exemplified based on the received sequence of signals x_{k}(t).
The 1^{st }image signal that is extracted according to the 2^{nd }method of the invention, is based on the reverberation corrected scattering signal obtained by band pass filtering the received sequence in the slow time domain around ω_{prf}/2, for example as indicated by the band pass filter 610 in
The amplitudes of these signals increase monotonously with τ. By choosing adequate frequency of the low frequency field (e.g. ω_{0}˜ω_{1}/10), the absorption attenuation of the low frequency field will be very low within the image range, and vary little between different objects and individuals at defined depth ranges. We can hence design the low frequency field and pulse amplitude so that we get a monotone increase in τ(t) as in
The filter outputs will also contain the nonlinearly scattered components of lines 602 and 603 as
which is maximum for τ=0, and decreases with depth as □ increases. However, the nonlinear scattering signal component in Eq. (15) will be negligible compared to the linear scattering signal component in Eq. (14).
The signals after the band pass filter can be used for further Doppler processing to produce Doppler spectra and radial Doppler image lines of scatterer velocities according to known methods, where the full 2D or 3D image then is generated by lateral sweeping of the beam. This method is particularly useful for Doppler estimation of myocardial displacements and displacement strain (radial gradient of displacement), as the reverberation noise strongly interferes with such estimations. For fluid velocity measurements, one should note that the object clutter signal in Eq. (14) is found around (ω_{prf}/2, and to suppress the object clutter to estimate the fluid signal, one can either use a band stop filter in slow time around ω_{prf}/2, or frequency mix the signal from ω_{prf}/2 to ω=0, and use standard, high pass type clutter filtering before estimating Doppler frequencies of the fluid signal.
One can also use slightly different and overlapping beam directions for each k, for example as obtained with a continuous sweep of the beam with an annular array. One might then also use an IIR filter for slow time filtering (band pass, low pass, high pass), where a low pass filter is illustrated in Eq. (77) for similar processing.
A first signal according to the 2^{nd }method of the invention to be used for a radial image line for the strength of the linear back scattering with suppression of the pulse reverberations, can hence be obtained as the envelope of one of the {tilde over (z)}_{lk}(t) of Eq. (14), or the average envelopes of the {tilde over (z)}_{lk}(t) for several k's. One can also form a linear combination of a set of received high frequency signals x_{k}(t), for example similar to Eq. (19) without delay corrections, that attenuates slow time frequency components around ω=0 while passing through slow time frequency components around ω_{prf}/2, to produce the signal z_{1}(t) as a reverberation corrected linear scattering signal for the radial image line determined by the beam direction, and form the envelope a_{1}(r) as
where Env{ } is the envelope operator and r=ct/2 is the depth range along the beam, and the formula for z_{1}(t) is an example band pass combination as in Eq. (19). The full 2D or 3D image is then obtained by lateral scanning of the beams.
The pulse reverberations are reduced by the accumulative delay effect of the nonlinear propagation velocity manipulation by the pressure, the same effect that forms harmonic distortion in the forward propagating pulse that is utilized in harmonic imaging. However, with the current method the nonlinear propagation is produced by the low frequency pulse which has so low absorption that the sensitivity with the method is similar to that for 1^{st }harmonic imaging. This allows the use of higher imaging frequencies than with 2^{nd }harmonic imaging with improved resolution, and particularly allows better imaging at deep ranges in dense objects like the liver, the kidneys, and the breast.
As the transmit beam with this method is a 1^{st }harmonic beam, it is easier to make a broader transmit beam with this method than with a 2^{nd }harmonic transmit beam. This allows use of more parallel receive beams to increase frame rate with 3D acoustic imaging.
To separate the nonlinear scattering components from the linear scattering components, one must delay correct (time shift compensate) the received signals, so that the frequency shift mixing of the switching nonlinear propagation delays disappears for the linear signal and line 602 disappears to leave only the nonlinear scattering line 603 in the band around ω_{prf}/2. The delay correction will depend on the amplitudes and/or the phases of the low frequency pulses relative to the high frequency pulses, and also varies with depth according to Eq. (10), and as exemplified in
According to one aspect of the invention, one can estimate delay corrections τ_{k}(t) by maximizing the power in the signal
Generally there is no reference signal, so for determining the delay corrections one must use one of the signals as reference, where the delay correction for this signal becomes zero. We hence can determine only K−1 independent delay corrections according to the methods discussed below. The summation represents a low pass filter in the slow time domain with the frequency transfer function
One can also use other variations of low pass filters, for example illustrated as 611 in
The delay corrections τ_{k}(t) introduce a frequency mixing, which move all the linearly scattered power to the line 621 in
The nonlinearly scattered signal can hence be obtained after delay corrections as
where the summation represents a band-pass filter around ω_{prf}/2 which for example with h_{k}(t)=1 takes the form
where other values for h_{k}(t) produces modifications to this filter, for example to the filter indicated as 610 in
The linear combination in Eqs. (17,19) give one signal output in the slow time domain, where it is clear to anyone skilled in the art to modify the equations as FIR or IIR filters that produces a set of output signals in the slow time domain similar as to discussed in relation to Eqs. (14, 16). Such sets of signals in the slow time domain would be used for further Doppler processing of the signals in the slow time domain, to form Doppler velocity or displacement strain of strain rate images of the object scatterers and fluid according to known methods.
To estimate the delay corrections we can according to one embodiment of the invention, divide the received time/depth interval T into shorter time intervals {T_{i}, i=1, . . . , l} that possibly overlap so that T≦Σ_{i}T_{i}, and we estimate optimal delay corrections for each interval T_{i }separately. The power of z_{1}(t) in T_{i }is then given by the functional
and the delay corrections are estimated by maximizing J_{1i }with respect to τ_{ip}. Examples of maximizing procedures are given below.
The nonlinear self distortion of the high frequency pulse described in
Examples of minimizing procedures are given in Eqs. (55-58). The variation in h_{k }also take care of variations in the transmit amplitude of the high frequency components when the amplitude and/or polarity and/or phase and/or frequency of the low frequency pulse change between transmitted pulses. The delays τ_{ip }are then efficiently obtained by maximizing J_{1i}, while the h_{k}'s are obtained by minimizing J_{ni}.
The maximization of J_{1i }estimates an average delay correction for each interval T_{i}. For best correction according to Eq. (17), one should assign these delay estimates to a point inside the corresponding intervals T_{i}, and produce an interpolated delay correction estimates τ_{k}(t) at each sample point of the fast time t between these selected points. The selected points can for example be the center of the intervals or the point of gravity of the power in the received signals in the corresponding intervals, or similar. Several methods of interpolation can be used, such as linear interpolation, spline interpolation to any degree, and Fourier interpolation.
The linearly scattered signal from the object in the lines 605 and 606 of
With reference to the discussion in relation to
Below, we shall also show a 3^{rd}, Eqs. (42-44), and a 4^{th}, Eqs. (45-48), method according to the invention where the effect of the pulse reverberation noise is suppressed directly in the estimation of the propagation delays. Before we discuss these methods we shall see how we can extract multiple image parameters signals with the 2^{nd}-4^{th }method of the invention.
The maximization of J_{1i }with z_{1c }as presented in Eq. (17), introduces delay corrections for both the nonlinear propagation time lag and the average Doppler time lag produced by object/transducer displacement in the interval T_{i}. The reason for this is that the summation filter as in Eqs. (17, 18) has a slow time frequency response with a maximum for ω=0, and the maximum of J_{1i }is found when all the linear energy is shifted to centered around ω=0. With a slow time low pass filter with a flat response around ω=0, and strong attenuation around ω_{prf}/2, one could have a maximum of J_{1i }that yields only the nonlinear propagation delays where the linearly scattered energy of lines 602 and 603 is moved to the filter pass band, without correction for the Doppler delay. However, such a filter is difficult to make with the limited number of slow time pulses used for each radial image line, and hence we get most robust estimation by using a low pass filter with a defined maximum at ω=0. The delay correction estimation then represents total propagation delay as the sum of the nonlinear propagation delay and average Doppler displacement delays (i.e. Doppler delay) of the relative movement between object scatterers and transducer.
As noted in relation to Eq. (17) we can estimate K−1 nonlinear propagation delays, i.e. one less than the number of signals. With three or more signals, accurate estimation of the delay corrections for the signal modeled in Eqs. (11-13) and a filter with maximal frequency response at ω=0, then gives the error free delay corrections as the total propagation delays which are the sum of the nonlinear propagation delays and the Doppler delays
τ_{k}(t)=(−1)^{k}τ(t)+2ν_{r}(t)T _{prf} k/c=(−1)^{k}τ(t)+kτ _{d} (23)
where τ_{d}=2ν_{r}(t)T_{prf}/c is the Doppler displacement delay due to radial scatterer displacement ν_{r}(t)T_{prf }between transmitted pulse complexes. The Doppler phase shift and Doppler frequency can be found as
This Doppler estimate is interesting to determine the radial displacement (from the phase in Eq. (24a) and velocity (from the angular frequency in Eq. (24b) of objects, for example the myocardium, as well as the radial displacement strain and strain rate as the radial gradient of the radial displacement and scatterer velocities. To estimate the Doppler shifts of scatterers in clutter noise, like blood or other fluids, one would first use a clutter high pass filter before the Doppler estimations, as described in relation to
The nonlinear propagation delay is found as
As the estimation of τ_{ik }contains errors, one can reduce the estimation error by averaging Eqs. (24,25) over neighboring k. With lateral movements of the scatterers or a fast mechanical sweep of the acoustic beam, there might be an inherent variation of τ_{i,k }with k due to exchange of the object structures in the beam from pulse to pulse, where one should limit number of pulses (k) to average over.
As p_{0}(r) can be determined from apriori measurement due to the low absorption of the low frequency pulse, one can from the nonlinear propagation delays estimate a 1^{st }quantitative nonlinear imaging parameter, representing the nonlinear forward propagation properties of the material. The increment in the delay corrections between neighboring intervals T_{i }represents a nonlinear forward propagation parameter that can be written as
where β_{nia }and κ_{ia }are spatial averages over the range interval corresponding to T_{i}, and p_{0ik }is the average amplitude of the low frequency component in the same range interval corresponding to transmit pulse no k. The 1^{st }quantitative nonlinear image parameter/signal (nonlinear propagation image parameter) is then obtained from Eq. (26) as
A 2^{nd }image signal to be used for imaging according to the invention, is the envelope a_{nc}(r) of the nonlinearly scattered signal z_{nc}(2r/c) of Eq. (19). This envelope is related to the nonlinear scattering parameters of the material as
where υ_{n}(r) is bandpass filtered around 2k_{1 }as discussed in relation to Eqs. (2,9) and averaged laterally together with the amplitude of the low frequency pulse p_{0 }across the high frequency beam profile for range r. The exponential term describes the absorption attenuation of the high frequency acoustic pulse in the object, and is compensated for by the user adjustable time/depth gain compensation G(r) in the acoustic instrument.
The absorption factor can be found by comparing a_{nc}(r) with the envelope of a 3^{rd }image signal, the linearly scattered signal after delay corrections, z_{1c}(t) of Eq. (17), which is related to the linear scattering parameters and the acoustic absorption in the object as
where υ_{1}(r) is bandpass filtered around 2k_{1 }and averaged laterally across the high frequency beam profile for range r as discussed above.
When p_{0}(r) is known, for example through calculations or measurements as described above, we can combine the signals in Eqs. (28,29) to obtain a 2^{nd }quantitative nonlinear image parameter/signal, the quantitative nonlinear scattering parameter/signal, of the object as
This adjusted time shift then holds the propagation information of β_{nia}κ_{ia }averaged over the local interval T_{i}, whereas the nonlinear scattering signal z_{nc}(t) holds information of local, spatial fluctuations in β_{n}κ in the interval T_{i}. One can also use the interpolated values of τ_{ik }along the fast time to τ_{k}(t), and let Eqs. (26,27) represent differentiation along the fast time samples to present a smoother version np(t) of the nonlinear image parameter np_{i}. Similarly, one can assign the values of np_{i }to points inside the intervals T_{i }and interpolate the values between these points for presentation of the image, similar to described for interpolation of the propagation delay above.
The methods of reducing the pulse reverberations as described in relation to
Writing the spatial variation of the acoustic propagation velocity as c_{0a}(r)=c_{0}+Δc_{0a}(r), where c_{0 }is the constant propagation velocity assumed (˜1.54 mm/μsec) when calculating the beam former delays according to assumptions of a homogeneous material, we can approximate the wave front aberration delay as
where r is the element position vector on the array surface of the actual element or sub-aperture, r _{f }is the position vector of the beam focus, and Γ(r,r _{f}) is the ray path from the element center r to the focus r _{f}. To correct for the wave front aberrations, one should delay the transmit pulses for the transmit beam and the receive signals for the receive beam for the elements at location r with a correction delay τ_{cor}(r)=−τ_{ab}(r), as described in U.S. Pat. No. 6,485,423.
There is a strong correlation between variations of mass density and compressibility between materials, which according to Eq. (3) implies that there is a correlation between variations in the compressibility and variations in the propagation velocity as
where β_{c }is a proportionality constant that is determined from experiments and we note that β_{na}κ_{a}=np is the quantitative nonlinear propagation parameter in Eq. (27), now estimated for each element or sub-aperture signal defined by the location r on the array surface. β_{n0}κ_{0 }and np_{avg }are the spatial average parameter for all elements and delay intervals. Inserting this expression into Eq. (31), we obtain an approximate estimate for the aberration corrections as
the last sum is independent of r. This estimate can also be used as a starting estimate for the procedures given in the above cited patents and patent applications.
The variation of np_{i}(r) with r is mainly produced by the propagation through the body wall, and as it is the variation of τ_{ab}(r) with element or sub-aperture location r that produces the aberrations, one gets good results by summing for intervals i in Eq. (33) slightly past the body wall only. In this near field region it is possible to design the low frequency field so that p_{0}(s) is approximately constant in the body wall for each element, so that it can be taken outside the integral for τ(r) in Eq. (10). We then see that we can relate the aberration correction delays directly to the nonlinear propagation delays past the body wall, and subtracted the spatial average of τ(r), τ_{avg}, across all the elements, i.e.
where p_{avg}(r) is the spatial average of the low frequency field p_{0}(s) along the propagation path Γ(r,R_{b}) from element location r through the body wall with thickness R_{b}. As noted above, the nonlinear propagation delays could with the 2^{nd }method be estimated from the 2^{nd }harmonic band of the received signals to reduce errors produced by pulse reverberation noise. In the 3^{rd }and 4^{th }method according to the invention presented below, the pulse reverberation noise is avoided directly in the processing, while in the practical situations the use of the 2^{nd }harmonic band will improve the accuracy with these methods.
From Eq. (8) we see that we can estimate the local absorption coefficient α(r) from the differential of the center frequency of the received high frequency signal as a function of range r. However, without pulse compression the down sliding of the frequency can make the receiver transducer limit the bandwidth of the received signal. Therefore, by using the pulse compression in Eq. (7) so that it balances the absorption down sliding of the frequency, we can place the received signal in the middle of the receive transducer band to avoid down-cutting of the signal bandwidth by the receive transducer band limitations, making it possible to estimate the local power absorption from measurements of the received high center frequency and the nonlinear propagation delays τ(r) as given in Eq. (10). One estimates τ(r) for example according to the 2^{nd }method above, or the 3^{rd }and 4^{th }method below, in addition to a transmission of a pulse complex as in
With reference to the discussion in relation to Eqs. (7,8) we get in this last situation a compression of the high frequency pulse length as dλ_{1}=−β_{na}κ_{0a}p_{00}k_{00}λ_{10}p_{0}(r)/p_{00 }dr over a propagation distance dr, where p_{0}(r) is the local amplitude of the low frequency pulse including variations from beam convergence/divergence and small absorption, p_{00 }and k_{00 }are the low frequency pulse amplitude and wave number at the array surface, and λ_{10 }is the high frequency wave length at the array surface. The received center frequency of the high frequency signal is related to the wave length as f_{1}(r)=c_{0a}/λ_{1}(r), which gives df_{1}=−f_{1} ^{2}/c_{0a}dλ_{1}−0.36B_{1} ^{2}(r)α(r)dr=β_{na}κ_{0a}k_{00}λ_{10}p_{0}(r)f_{1} ^{2 }dr/c_{0a}−0.36B_{1} ^{2}(r)α(r) dr. The pulse bandwidth is related to the pulse length R_{1 }as B_{1}=c_{0a}/R_{1}. Per the discussion above, the pulse length compresses over a propagation distance dr as dR_{1}=−β_{na}κ_{0a}p_{00}k_{00}R_{10}p_{0}(r)/p_{00 }dr where R_{10 }is the pulse length at the array surface. The absorption down conversion preserves the signal bandwidth which gives dB_{1}=−B_{1} ^{2}/c_{0a}dR_{1}=β_{na}κ_{0a}k_{00}R_{10}p_{0}(r)B_{1} ^{2 }dr/c_{0a}. This analysis gives the following differential equations for f_{1 }and B_{1}
Eq. (35b) can be integrated directly, which allows us to estimate the local absorption coefficient from the measured nonlinear propagation delay τ(r), its gradient, and the gradient of the center high frequency f_{1}(r) as
where τ(r) is given in Eq. (10) and we have used that dτ/dr=−β_{na}(r)κ_{0a}(r)p_{0}(r)/c_{0a}(r) and that R_{10}B_{10}=c_{0a}(0), and k_{00}c_{0a}(0)=ω_{0}, the angular center frequency of the low frequency pulse. By adjusting the amplitude of the low frequency pulse one can avoid that the received high frequency signal is modified by the edges of the high frequency receive band so that Eq. (35b) is valid which allows the integration to Eq. (36b) which further allows the estimation of □(r) from Eqs. (35a,36a).
The nonlinear parameter β_{n }becomes very low for hard materials as does the compressibility κ. Therefore, in particular, at interfaces between soft and harder materials, for example tissues with high density of connective fiber molecules, calcifications, or other high density materials, the nonlinear scattering becomes strong. Similarly do one get strong nonlinear scattering at interfaces to softer materials such as fat, foam cells, and especially micro gas bubbles in the tissue where the nonlinear scattering is further enhanced as described below. The same is true for the interface between hard rock and fluid or gas in geological structures, swim bladder of fish or lungs of sea animals in water, mines in soil or on a soft sea bed, etc. The nonlinear imaging hence enhances the visualization of such structures. The invention is therefore useful to visualize micro-calcifications in soft tissue, for example for imaging of tumors in the breast and other tissues, or atherosclerotic plaque, which is difficult to visualize with current acoustic imaging methods. Also, with less dramatic changes in material compressibility, as the compliance decrease with in-growth of connective tissue, or compliance increase with in-growth of fat or foam cells, the nonlinear parameters estimated with these methods give increased image contrast for the tissue changes, compared to current imaging. The image parameters in Eqs. (27,30) then allows for quantitative assessment of the tissue changes.
For gas bubbles, either found naturally in the object as hydrocarbon gas or diver decompression bubbles, or injected into the object as ultrasound contrast agent, the bubble scattering dynamics is described by a differential equation, providing a resonant scattering with a frequency dependent phase lag between the incident and the scattered wave, contrary to scattering from fluid or solid objects where the frequency variation of this phase lag is very low. The low frequency pulse manipulates the micro-bubble diameter (small diameter with positive p_{0}, and large diameter with negative p_{0}), and hence the micro-bubble resonance frequency. This manipulates the phase lag of the scattered signal for the high frequency pulse, in addition to the amplitude of the scattered signal. The manipulation is particularly strong for high frequency pulses in the neighborhood of the micro-bubble resonance frequency as described in U.S. patent application Ser. No. 10/851,820 filed May 21, 2004.
The corrected nonlinear signal z_{nc}(t) for example according to Eq. (19) then contains close to all of the high frequency scattered power from the contrast agent bubbles (both linear and nonlinear components). For a scattering object that contains micro gas bubbles, the present invention therefore significantly increases the CNR (Contrast to Noise Ratio) relative to existing methods of imaging of such bubbles by extracting close to the total scattered high frequency signal power from the micro-bubbles, in particular the strong linear components and not only nonlinear components. Corrections for the low frequency pulse switching of the nonlinear propagation delays provide a suppression of the linearly scattered power from the object providing a large COR (Contrast to Object Ratio). The method hence separates nonlinear forward propagation effects from local, nonlinear scattering and utilizes the local manipulation of the frequency variation of the phase of the scattered signal from gas bubbles to obtain strong, local signal from gas bubbles with strong suppression of the local object signal.
This is different from current methods of contrast agent imaging, where the nonlinear propagation produces an accumulated effect on the forward propagating pulse that also enhances the linear scattering from the tissue in the detection process, so that this linear signal from tissue masks the signal from micro-bubbles (and also the nonlinearly scattered signal from the tissue). Similar effects are found with detection of gas bubbles past a cloud of gas bubbles in geologic structures, and objects past a school of fish with swim bladders or sea animals with lungs.
When the pulse passes through a cloud of gas bubbles (also swim bladders), these will provide increased, accumulative nonlinear propagation lag on the forward propagating pulse, and also nonlinear variation in the pulse amplitude, a phenomenon that increases the need to correct for the nonlinear propagation delays and variations in pulse amplitude to obtain good suppression of the linearly scattered object signal beyond the cloud. The current invention therefore has strong advantages above known methods of micro-bubble imaging. For example, with harmonic imaging the increased, accumulative harmonic distortion for a pulse that passes through a cloud of micro-bubbles is found as strong harmonic components in the linear scattering from tissue beyond the cloud. This for example can provide strong harmonic scattering from the myocardium for a pulse that has passed through a cloud of contrast agent in the ventricle, masking the scattered signal from contrast agent micro-bubbles in the myocardium. This effect can falsely indicate blood perfusion in a region of myocardium with very low or no perfusion, and also indicate gas in a geological structure without gas. With the current method, the effect of a cloud of micro-bubbles in the ventricle on the forward propagating pulse is removed for scattering from the myocardium past the cloud by the corrections for the nonlinear propagation delays. The invention separates the local nonlinear scattering from the accumulative nonlinear forward propagation effect, and hence safeguards that one measures the local nonlinear scattering that greatly prevents such false indications of non-existing micro-bubbles in the myocardium.
Relative to nonlinear harmonic imaging methods, the present invention can use a more broadband transmit pulse and will hence achieve a higher image range resolution. In addition, a higher imaging frequency can be used, resulting in a significant increase in both lateral and range resolution relative to other methods of imaging gas bubbles. The performance of the present invention is less sensitive to the amplitude of the imaging pulses compared to harmonic imaging methods. Together with the indicated suppression of received linearly scattered signal with resulting increase in COR, this facilitates high resolution non-destructive detection and imaging of single contrast agent bubbles with low Mechanical Index (MI).
In medical applications, the improved sensitivity and high resolution imaging of ultrasound contrast agent has strong potentials in imaging of changes in micro-vasculature, for example neo-angiogenesis or necrosis in tumors, or reduced blood perfusion in the myocardium where some standard methods of using inflow time of contrast agent has been developed. The quantitative parameters in Eqs. (27,30) provide quantitative information on the contrast agent density in the tissue, and hence provide an improved assessment of the relative volume of the micro vasculature. The methods also have applications to estimation of relative gas volume in geological structures, and density and size of fish with swim bladder or sea animals with lungs. By destroying the contrast agent bubbles in a region and measuring the inflow time, one can obtain quantitative values for blood perfusion through the tissue, according to known principles. The high sensitivity, high resolution imaging of contrast agent is also useful for tracing of lymphatic drainage to find sentinel lymph nodes in tumor surgery.
As Eqs. (27,30) give imaging parameters that do not depend on absorption in the object, one can use these object parameters to characterize the object (for example fat or connective tissue content in soft tissue, content of oil, gas or water in porous rock, or amount and size of fish and sea animals). In addition one can determine the local variation of the propagation velocity with temperature based on experiments to monitor changes in tissue temperature with thermal treatment of diseased tissue, for example high intensity focused ultrasound (HIFU), RF ablation, or cryo-surgery. The temperature can be monitored from changes in the quantitative parameters, but also from changes in the propagation velocity which causes time lags between the back scattered signals from image to image as the temperature is changing. Radial gradients in this time lag determines the local temperature.
As the local linear scattering of the high frequency pulse is not influenced by the low frequency pulse, it is implied that variations of the amplitude and/or the phase and/or frequency of the low frequency components between transmitted pulses other than that shown in
The low and the high frequency bands of the transmitted pulses are often so widely separated that one can prefer to use separate transducer arrays to transmit the two bands of the pulse. Such arrays can be made as concentric rings with different resonant frequencies, where the beams from the arrays automatically overlap, or the arrays can be mounted by the side of each other with skewed crossings of the beams.
When we use two different arrays for the low and the high frequency components, with beams that overlap in a skewed direction, the phase relationship between the low and the high frequency pulses can have a strong spatial dependency, depending on the geometric arrangement and the dimensions of the two radiating array surfaces. An example arrangement according to the invention of separate low and high frequency transducer arrays radiating along the z-axis is shown in
The boundaries of an example low frequency beam are shown as 803, whereas the boundaries of the high frequency beam are indicated as 804. We note that we have a near-region 805 in front of the array where there is limited overlap between the low and the high frequency beams, hence providing particularly strong suppression of the body wall reverberations of the outgoing pulse with the methods described in relation to Eqs. (14,16) and
We further note that the propagation distance from the low frequency array to a scatterer close to the z-axis is larger for the low frequency pulse than the high frequency pulse, depending on the geometric dimensions of the low and high frequency arrays. The axial low frequency field at the point 806 at z and angular frequency ω_{0}=ck_{0 }is
where P_{lt }is the transmit pressure on the array surface, R_{lo}(z) is the distance 807 from the outer edge of the low frequency array to 806 on the z-axis and R_{li}(z) is the distance 808 from the inner edge of the low frequency array to 806. The phase term represents the average propagation lag
where τ_{l}(z) is the propagation lag from the low frequency array to 806 and τ_{h}(z) is the propagation lag from the high frequency array to 806, where R_{ho}(z) is the distance 809 from the outer edge of the high frequency array to 806. In addition to the propagation phase lag, one will encounter changes in the sign of the sine term in Eq. (37) as part of the phase of H_{l}(z) and is found as steps of ±π when a zero in the sine term is passed. As 810 in
Δθ_{l}(z)=−∠{H _{l}(z)}−ω_{0}τ_{h}(z) (39)
as a function of z for a typical geometry with a high frequency aperture D_{h}=7 mm, with inner and outer parts of the low frequency aperture D_{li}=10 mm and D_{lo}=15 mm. A drop in the phase lag of n is found at 811 (z˜6.6 mm) as the difference propagation phase Δθ_{l}(z)=ω_{0}(R_{lo}(z)−R_{li}(z))/2c_{0 }between the outer and inner boarders of the low frequency array passes zero, which produces a change from −1 to +1 in the sign of the sin term in Eq. (37). The amplitude of low frequency field H_{l}(z) is shown in un-scaled log values as 812, and we note that a zero in the amplitude coincides with the π-shift in the phase. Zeros in the field are found when the difference in the phase propagation lag between the outer and inner edges of the array to the field point is an odd number of π, with a production of a step π in Δθ_{l}(z).
For comparison is also shown as 813 the difference in phase propagation between the low frequency and the high frequency arrays, given as
Δφ_{l}(z)=ω_{0}(τ_{l}(z)−τ_{h}(z)) (40)
We note that Δθ_{l}(z) follows Δφ_{l}(z) before the −π step at 811 and follows with a difference of π thereafter. Due to the large and z-dependent propagation phase lag between the low and the high frequency arrays, one will get a z-dependent relative position between the high and low frequency pulses. For example, a high frequency pulse that originally starts at the top ridge of the low frequency pulse shown as 814, slides towards the bottom trough of the low frequency pulse at 815 when the phase lag Δθ_{l}(z) has changed π, and so forth.
Inserting a pressure
p _{0}(s)=P _{lt} |H _{l}(s)|cos Δθ_{l}(s) (41)
into Eq. (10) with β_{n}=5, κ_{0a}=400·10^{−12 }Pa^{−1}, and c_{0a}=1540 m/s, we get with P_{lt}=50 kPa a nonlinear propagation lag shown as 816 shown in
For imaging of contrast agent, one would generally use higher low frequency pressure p_{0}, say p_{0}˜200 kPa, which would give τ_{max}=8 nsec in the above example and reduce the suppression of the linearly scattered high frequency signal by 2 sin(ω_{l}τ_{max})˜−6 dB for f_{1}=10 MHz, even with the large sliding of the phase relation between the low and the high frequency components as found in this design example. Hence, an amplitude of 50 kPa for the low frequency pressure is very low, but can provide interesting imaging of contrast agent and other micro gas bubbles in special situations. It is hence possible to enhance contrast agent signal to some degree without corrections for the nonlinear propagation delays, provided the arrays are designed so that an oscillatory variation of the nonlinear propagation delays as in 816 is found. We should note that when the high frequency pulse is close to zero in the low frequency oscillation, the nonlinear scattering and forward propagation effect is low. This effect can be avoided by using different phases between the low and high frequency components in consecutive transmit pulses, which would shift the spatial location where the high frequency pulse is found close to a zero in the low frequency oscillation. Zeros in the amplitude |H_{l}(s)| of the low frequency field can be shifted in space by changes the low frequency center between consecutive transmit pulses. Also, due to the width of the high frequency beam, one will also pick up signal from outside the axis where |H_{l}|≠0, and the pulse is composed of a frequency band which averages zero points for many frequencies. The amplitude zeros can be avoided by reducing the width of the low frequency array, which would however also lower the pressure amplitude to drive voltage ratio for the array.
However, to extract nonlinear scattering from other objects, like from micro-calcifications in soft tissue and gas in porous rock, and produce quantitative object parameters, it is necessary to use higher low frequency amplitudes, where it becomes mandatory to correct for the nonlinear propagation delays in order to suppress the linear scattering and extract the nonlinear scattering. This is also the case for utilizing the method to suppress pulse reverberation noise as discussed in relation to
Often one would select a design of the low and high frequency radiation apertures so that one gets minimal delay sliding between the low and high frequency pulses, to maximize the nonlinear manipulation of the scattering and propagation parameters for the high frequency pulse along the whole image range. This is necessary to produce a monotone increase of the nonlinear propagation delay as shown in
To reduce the effect of the pulse reverberation noise on the estimation of the corrections for the nonlinear propagation delays, as well as in the nonlinearly scattered signal, one can in a 3^{rd }method according to the invention transmit more than two pulses with more than two different amplitudes and/or phases and/or frequencies of the low frequency pulse. As a first example according to the invention where there is no movement between the scatterers and the transducer array (i.e. zero Doppler displacement), we for example transmit low frequency pulses with amplitudes +p_{0}, 0, −p_{0 }where the high frequency pulse follows close to the peak or trough of the low frequency pulse as for example shown in
y _{1}(t)=x _{l}(t+τ _{i})+x _{n}(t+τ _{i})+r(t)
y _{2}(t)=x _{l}(t)+r(t)
y _{3}(t)=x _{l}(t−τ _{i})−x _{n}(t−τ _{i})+r(t) (42)
where x_{l}(t) is the linearly scattered signal, x_{n}(t) is the nonlinearly scattered signal with amplitude +p_{0 }of the low frequency pulse, r(t) is the reverberation signal for the high frequency pulse, and −τ_{i }is the nonlinear propagation delay for the high frequency pulse when it propagates on the positive ridge with pressure +p_{0 }of the low frequency pulse. With one embodiment according to the invention, we form combinations of these three signals as
z _{1}(t)=y _{2}(t)−y _{1}(t)=x _{L}(t,τ _{i})−x _{n}(t+τ _{i})
z _{2}(t)=y _{3}(t)−y _{2}(t)=x _{L}(t−τ _{i},τ_{i})−x _{n}(t−τ _{i}) (43)
where x_{L}(t,τ_{i})=x_{l}(t)−x_{l}(t+τ_{i}) is a combination of the linearly scattered signals. The combinations have removed the pulse reverberation noise from z_{1}(t) and z_{2}(t), and as the amplitude of the nonlinearly scattered signal is much lower than for the linearly scattered signal, we have approximately z_{2}(t)=z_{1}(t−τ_{i}), and we can for the interval T_{i }determine the delay correction τ_{i }by maximizing the following functional
with respect to τ_{i}.
When τ_{i }is estimated, we can solve the linear and the nonlinear scattering signal components from Eq. (43), for example through Fourier transforming Eq. (43) in the fast time that produces a set of linear equations in X_{L}(ω) and X_{n}(ω).
The essence of the above procedure is that three or more pulses with three or more levels of manipulation of the nonlinear propagation delay allows us to combine the measurements to provide at least two signals where the pulse reverberation noise is highly suppressed, and these new signal can be used for estimation of the delay corrections. With the procedure as described, the difference between the amplitudes of neighboring low frequency pulses must be constant, giving a nonlinear propagation delay for each signal that is a whole multiple of one delay. With a constant radial movement between the transducer array and the scatterers that produces a constant Doppler delay between the received signals for each pulse, we have a fourth unknown to estimate, which requires that we transmit at least four pulses with different amplitudes of the low frequency pulse, or one can use 5 pulses with the 3 amplitudes of the low frequency pulse as in Eq. (42), where by maximization of a signal power in the same way as in Eqs. (21,44) we estimate combined nonlinear propagation and Doppler delays. The nonlinear propagation and Doppler delays can then be separated in operations like in Eqs. (24,25) where the details to such and similar procedures can be worked out according to the invention by anyone skilled in the art.
However, τ_{i }itself must be estimated from combinations like Eq. (44), or the equivalent in the Fourier transform, and as z_{1 }and z_{2 }contains both the linear and the nonlinear scattering signal, the nonlinear scattering signal will introduce an error in the delay correction estimate, albeit very low, that in turn introduces an error in estimation of the nonlinear scattering signal, in the same way as the estimations given in Eqs. (17-22).
The fundamental reason for this error is that we for stationary objects have four unknowns: The linear scattering x_{l}(t), the nonlinear scattering x_{n}(t), the pulse reverberation noise r(t) and the nonlinear forward propagation delay τ(t) (Stationary object with no Doppler delay). As τ(t) has a slow variation with t, we can approximate it as constant over the time interval T_{i }as above. For most accurate estimation of all four unknown, one should have at least 4 measurements, for example with the four levels +p_{0}, +p_{0}/3, −p_{0}/3, −p_{0 }of the low frequency pulse to give
One could then eliminate r(t) as in Eq. (43) to produce 3 signals z_{1}(t), z_{2}(t), and z_{3}(t), and determine τ_{i }to align all these signals with maximization of a functional similar to Eq. (44). The influence of x_{n}(t) in the estimate of τ_{i }will be lower than with 3 measurements, but the error it introduces is not completely eliminated.
A better approach according to a 4^{th }method of the invention, that decouples the influence of x_{n}(t) on the determination of τ_{i }is found through the Fourier transform of Eq. (45) over the interval T_{i }where τ(t) can be approximated by the constant τ_{i}, which gives an over-determined set of linear Equations of the form
Y (ω)=A(ωτ_{i}) X (ω) (46)
where Y(ω)^{T}={Y_{1}(ω), Y_{2}(ω), Y_{3}(ω), Y_{4}(ω)}, X(ω)^{T}={X_{l}(ω), X_{n}(ω), R(ω)}, and A(ωτ_{i}) is a 4×3 dimensional matrix obtained from the Fourier transform of Eq. (43) according to known methods. With such an over-determined system, one can find the X(ω) that provides the best approximation in the least square sense using the pseudo-inverse of the matrix A(ωτ_{i}) as
{tilde over (X)} (ω)=(A ^{H} A)^{−1} A ^{H} Y (ω) (47)
where A^{H}(ωτ_{i}) is the Hermittian transpose of A(ωτ_{i}). τ_{i }can now be estimated from a minimization of the error in the least square adaptation
with respect to τ_{i}, a process often referred to as total least squares. This method provides a systematic procedure to utilize at least four measurements with at least four different levels of the low frequency pulse to estimate all the four unknowns, especially the nonlinear propagation lag, with highly reduced influence from pulse reverberations and the nonlinear signal, while in many situations the methods described in relation to Eqs. (17-22 and 42-45) provide adequate results. We also note that the amplitude of the low frequency pulses can be set arbitrary so that the nonlinear delays for the different pulses can be set as τ_{i}=k_{i}τ, where k_{i }is ratio of the low frequency amplitudes to a reference and t is the nonlinear propagation delay with this reference that is estimated according to the procedure above.
We also emphasize that methods according to this structure provide estimates of the linearly and nonlinearly scattered signals (and also the pulse reverberation noise) with minimal influence from the pulse reverberation noise and in principle full decoupling between the linearly and nonlinearly scattered signal estimates. With this method one will hence use the estimated x_{l}(t) to represent the 1^{st }imaging signal of linear scattering with strong suppression of the pulse reverberation noise as Eq. (14) with the 2^{nd }method, and the 3^{rd }imaging signal of the linear scattering as in Eq. (29), while the estimated x_{n}(t) represents the 2^{nd }image signal from nonlinear sacttering as in Eqs. (19,28), and both are further used in the calculation of the 2^{nd }quantitative nonlinear scattering parameter in Eq. (30), where the estimated nonlinear delay t is used for estimation of the 1^{st }quantitative nonlinear parameter in Eq. (27).
We can as for the previous methods add a constant Doppler delay τ_{d }between the received signals for each transmitted pulse complex, which also gives us 5 unknown as X_{l}(ω), X_{n}(ω), R(ω), τ, and τ_{d}. These can be estimated from the received signals from 5 transmitted pulse complexes, for example with the amplitudes (+p_{0}, +p_{0}/2, 0, −p_{0}/2, −p_{0}) of low frequency pulses and the same amplitude of the high frequency pulse by minimization δY in Eq. (48) with.
The above discussion then gives a general outline on how to handle even more complex situations that produces more unknowns to be estimated. For example, we can have uncertainties in the amplitudes of the transmitted low frequency pulses, for example as +p_{0}, +a_{1}p_{0}/3, −a_{2}p_{0}/3, −a_{3}p_{0 }for Eq. (45) where the vector a=(a_{1}, a_{2}, a_{3}) represents unknown variations in the amplitudes. These uncertainties will both influence the amplitude of the nonlinear scattering and the delays in Eq. (45) and we see that we get 7 unknowns, i.e. X_{l}(ω), X_{n}(ω), R(ω), τ, a_{1}, a_{2}, and a_{3 }which means that we have too few measurement equations for adequate determination of τ in the minimization of δY in Eq. (48). We therefore must add new measurements with careful instrumentation so that the number of amplitude uncertainties is minimized. For example one can with careful instrumentation possibly have only one uncertain amplitude a with 5 transmit pulses and amplitudes of the low frequency pulse as (+p_{0}, +ap_{0}/2, 0, −ap_{0}/2, −p_{0}) which gives 5 unknowns X_{l}(ω), X_{n}(ω), R(ω), τ, a, i.e. the same number as equations where a minimization of δY in Eq. (48) will provide all 5 unknowns with no Doppler delay.
One should note that the 3^{rd }and 4^{th }method could as well as the 1^{st }and 2^{nd }method use the 2^{nd }harmonic band of the received signal for the processing and image signal formation, with the advantage of even better suppression of the pulse reverberation noise in the image signals, but at the cost of less image range for the same image frequencies. The instruments as discussed below therefore have the flexibility to select between the 1^{st }and 2^{nd }harmonic bands of the received signals for the processing according to this invention.
The nonlinear manipulation of the scattering and propagation parameters for the high frequency pulse by the low frequency pulse can produce favorable results also with tomographic computer image reconstruction from transmission and angular scattering measurements of the object, for example with an instrumentation as illustrated in
where Γ_{f}(2a;e _{i}) is the forward propagation path along the beam axis across the whole diameter 2a of the array. The propagation delay with zero amplitude of the low frequency pulse is t_{0f }while the nonlinear delay produced by the low frequency pulse is given by τ. The received signal at 907 will first be the transmitted pulse followed by a tail of multiple reflected pulses. However, with reference to
According to the methods of computer tomography reconstruction, one can use the delay without the low frequency pulse, t_{0f}, to find the propagation velocity c_{0a}(r) where r is the spatial coordinate in the plane. Adding a low frequency pulse and measuring the nonlinear propagation lag τ, then allows for reconstruction of β_{na}(r)κ_{0a}(r). Using the amplitude of the forward propagated pulse, one can also reconstruct the spatial variation of the absorption in the object. Moving the scan plane in the vertical direction then allows for 3D imaging of the object. All these reconstructed images will have a spatial resolution limited by the transmitted beam width, as the pulse delay and amplitude is an averaging over the whole beam, whereas the model in Eq. (49) assumes all delays confined to the beam axis. A similar model would be used for the absorption.
Improved resolution can be obtained by also using the angularly scattered signal, in methods referred to as diffraction tomography, reflective tomography, inverse scattering tomography, etc. On elements at an angular direction to the transmitted beam direction, for example 908 in
where we in the last expression for t_{0 }has assumed a constant propagation velocity c_{0a }along both Γ_{f }and Γ_{s}. As discussed in relation to
The delay corrections that maximizes functionals J_{1i }of the types shown in Eq. (21) and similarly Eq. (44) can (both for the 1^{st }and 2^{nd }harmonic bands of the received signals and also the complete received signals) for example be found from differentiation illustrated for Eq. (21) as
where the total number of delays are K−1 per the discussion in relation to Eq. (17). Due to the Hermittian symmetry of {circumflex over (R)}_{ki} ^{i}, this equation can be modified as
where Re denotes the real part of the sum. Noting that ∂x_{l}(t+τ_{ip})/∂τ_{ip}=∂x_{l}(t+τ_{ip})/∂t, we can rearrange Eq. (52) to
where {circumflex over (z)}_{li}(t) is the signal in Eq. (17) delay corrected with □_{ik}. As we do not have an analytic expression of {circumflex over (R)}_{kl} ^{i}, the delays that satisfy this set of equations must be found numerically, for example through the following iteration scheme
where q denotes the iteration step-number and μ is a gain factor that is adjusted to assure convergence of the iteration.
The minimization of J_{ni }in Eq. (22) with respect to h_{ip }can be done by equating the derivatives to zero, i.e.
Due to the Hermittian symmetry of {circumflex over (N)}, this equation transformed to the following eigenvector equation
The minimum of J_{ni }is then given by the eigenvector with the smallest eigenvalue λ_{i}. Introducing the expression for the correlation matrix in Eq. (22) allows rearrangement of Eq. (56) to
where {circumflex over (z)}_{ni}(t) is the signal in Eq. (19) delay corrected with τ_{ik}. Hence, the correction amplitude is found through correlation between the RF element signals and the delay and amplitude corrected high pass filter output RF signal where h_{ik }is involved. Eq. (57) can then be solved in an iterative procedure, for example as
We have in the above analysis used the complex analytic representation of the signal that can be obtained from the received RF signal by the use of the Hilbert transform as in Eq. (11). Approximations of the Hilbert transform is given in Eq. (74). It is also clear to anyone skilled in the art, that the above analysis can be modified to operate on the received RF-signals as the real part of the analytic signal.
For the procedures in Eqs. (54,58) it is natural to start the estimation for the interval T_{0 }at the lowest lag, followed by estimation in consecutive intervals. One then uses the estimated delay corrections τ_{ip }filter amplitudes h_{ip }for the preceding interval as the initial values in the iteration procedures for each new interval, starting with ρ_{0p,0}=0 and h_{ip}=1 as initial conditions for the first interval. An added advantage of using the estimated values for the preceding interval as initial conditions, is that when the step in the phase ω_{1}δτ_{ip }associated to the delay is less than ±π/ω_{1 }between each interval, one is able to track delays with associated phases |ω_{1}τ_{ip}|>>π over the whole depth range.
When the bandwidth of the signals is sufficiently low, and the delays are less than a period of the center frequency, the following approximations are adequate
where in Eq. (56b) we have split the delay correction as
τ_{ik}=δτ_{ik}+{tilde over (τ)}_{ik} (60)
where {tilde over (τ)}_{ik }is an estimate or approximation to τ_{ik }where we for example below will use {tilde over (τ)}_{i,k}=τ_{i−1,k}. Eq. (59b) hence represents a combination of phase delay with the phase ω_{l}δτ_{ik}, and true delay with {tilde over (τ)}_{ik }of the an estimate of τ_{ik}. Improved accuracy of this approximation is obtained by bandpass filtering the signals around □_{1 }to reduce the bandwidth of x^_{k}(t). Introducing the approximations of Eqs. (59) modifies Eqs. (53,54) as
Equating the last expression to zero, allows us to calculate δτ_{ip }as
where ∠{ } denotes the phase angle of the complex expression. This gives the following iterative equation for the amplitude and delay corrections
where one would typically start the iteration with τ_{ip,0}=τ_{i−1,p }with τ_{0p,0}=0 as initial conditions for the first interval.
With the approximation in Eq. (59), we can develop the maximization of the power in Eqs. (17,21) into an eigenvalue problem. We use the split of the delay as in Eq. (60), which gives
The functional in Eq. (21) is modified as
Maximization of J_{1i }with respect to the amplitude and phase of s_{ik}, gives s_{ik }as the solution to the following eigenvalue problem
{circumflex over (R)}^{i} s _{i}=λ_{i} s _{i }{circumflex over (R)}^{i}={{circumflex over (R)}^{i} _{km}}s _{i}={s_{ik}} (66)
for the largest eigenvalue λ_{i}. The maximum of J_{1i }is found by best possible alignment of the component signals so that optimal correction delays are found from the phases of the eigenvector components. The components of the s _{i }vectors defined in Eq. (64) all have unity module, while the components of the eigenvectors in Eq. (66) generally will have modules different from unity because the magnitude of J_{1i }can be further increased by putting different weights on different component signals. The phases of the eigenvector components will then give the delay corrections that maximizes the power in Eqs. (17,64). We note that the matrix {circumflex over (R)}^{i }is Hermittian, i.e. {circumflex over (R)}_{lm} ^{i}=(
The accuracy of the approximation in Eq. (59) and hence the eigenvector solution in Eq. (66) improves by reducing the bandwidth of the signals. The Fourier transform of the partial sum signals at ω_{1 }provides so strong band pass filtering that the differentiation in Eq. (59a) and the phase delay in Eq. (59b) becomes exact. However, to avoid that the phase delay exceeds the fundamental range of ±π, it is still advantageous to split the delays as in Eq. (60), which gives
One can then determine the delays by maximizing the functional
The maximization will lead to an eigenvalue problem for each frequency, similar to Eq. (66), where the delay corrections are found from the phases of s_{ik}. As the phase delay for the Fourier transform is an accurate representation of a true delay, Eq. (67) is fully accurate as opposed to Eq. (64) where the phase delay is taken only at the center frequency.
A bulk of the analysis above is based on availability of the complex analytic RF element signals and further processing of this to the corrected receive signal. The calculations can be simplified through the following exercises, where we note that the phase in Eq. (62) can be calculated as
The integration over T_{i }is an estimate of the following ensemble expectation values
∫_{T} _{ i }dtA_{i}(t)≈T_{i}
The approximation in Eq. (70) turns to equality as T_{i}→∞ where for finite T_{i }one will have a random estimation error of the ensemble averages, which is the reason for the approximation sign in Eq. (70). From signal processing textbooks we find that
For finite T_{i}, the random estimation errors produce the following approximations
We further note that
where the approximation of the Hilbert transform through delaying the signals π/2ω_{1 }is adequate for narrow band signals centered at ω_{1}. Combining Eqs. (73,74) we get the simplified expressions
In this expression, the Hilbert transform or its delay approximation operates only on the corrected high-pass filter output, and hence only have to be done on one signal, simplifying the operations. A similar expression could be developed from the last part of Eq. (73), but there the Hilbert transform or its delay approximation must be done on received signals x_{p}(t), which requires more processing. Some reduction in estimation variance can be obtained by combining the first and last expression in Eq. (73) along the lines of Eq. (69).
We also note the following relations
which allows that the above operations on the analytic signal can be substituted by operations on the complex envelope, where its real and imaginary parts are found as the in phase and the quadrature components of standard quadrature demodulator output signals as described above.
With a continuous, mechanical sweep of the acoustic beam it is also interesting to use Infinite Impulse Response (IIR) low-pass and high-pass filters, where an example 1^{st }order lowpass filter can be described as
where T is the filter time constant with cut-off frequency ω_{c}=1/T, the subscript k denotes the pulse number as before, and the image signal z_{ik}(t) is updated for each new transmit pulse. The same equation structure holds for the analytic signal and its complex envelope.
A recursive scheme of estimation of the correction delays τ_{ik }for each new transmitted beam is now developed. We assume that τ_{im }are given for m up to k−1. With the new transmitted beam no k with the received signal x_{k}(t) one then wants to estimate the delay corrections τ_{ik}. This estimation is then typically done through minimizing the functional where we have used the analytic signal approximation for simplicity
Differentiation with respect to τ_{ik }gives
Inserting the approximation in Eq. (59) we get
As the last term under the integral is real, equality of the above expression is found when
Higher order high pass filters can be described in a similar fashion with a vector state space representation of the filters known to anyone skilled in the art. A recursive estimation of τ_{ik }along the lines shown above can then be done by anyone skilled in the art.
For adequate suppression of the linearly scattered signal, one needs a delay accuracy that is much lower than the sampling interval of the received RF signal in most acoustic imaging system. This accuracy can be obtained through interpolation of the signal between the samples according to known methods. Less, but often adequate accuracy can be obtained by a combination of delay correction with the accuracy of the nearest sample
and a phase correction ω_{l}δτ_{ik }in analogy with Eq. (59) as
where δτ_{ik}=τ_{ik}−
The 1^{st }method according to the invention described in relation to
The delay corrected linear (1009) and nonlinear (1011) signals are then together with the estimated delay corrections (1006) fed to a quantitative parameter estimation unit 1012 that calculates one or more of the quantitative nonlinear propagation parameters 1015 according to Eq. (27) or similar, and the quantitative nonlinear scattering parameters 1016 according to Eq. (30) or similar, and the local acoustic absorption coefficient α(r) 1017 according to Eq. (36). The delay correction unit 1005 can also present Doppler phases 1013, for example according to Eq. (24), that are fed to a Doppler unit 1014 that calculates the radial scatterer displacement 1018, radial scatterer velocity 1019, radial scatterer displacement strain 1020, and radial scatterer displacement strain rate 1021, or other parameter calculations. The signals 1003, 1009, 1011, 1015, 1016, 1017, 1018, 1019, 1020, 1021 are then typically passed to further processing and displays to generate full acoustic images according to known methods.
The 4^{th }method according to the invention described in relation to Eqs. (45-48) operates more in a batch mode to estimate the corrections for the nonlinear propagation delays and the linearly and the nonlinearly scattered signal. Processors for such estimations can then be represented by the block diagram in
The processing units can then be implemented both for tomographic reconstruction methods based on transmission and angular scattering measurements as in
The high frequency part of the array can in full 3D imaging applications have a large number of elements, for example 3000-10,000, and the number of receive and transmit channels are then typically reduced in a sub-aperture unit 1202, where in receive mode the signals from several neighboring array elements are delayed and summed to sub-aperture signals 1203 for further processing. For aberration corrections, the widths on the array surface of the sub-aperture groups are less than the correlation length of the wave front aberrations, where a typical number of sub-aperture groups and signals could be 100-1000.
For transmission of the pulse complexes, the HF transmit beam former 1204 feeds pulses to the sub-aperture unit 1202, that delays and distributes the signals to all or sub-groups of HF-array elements, while the LF transmit beam former 1205 simultaneously feeds pulses to the LF array elements. The pulse complex transmission is triggered by the instrument controller 1206, which communicates with the sub-units over the instrument bus 1207.
The receive sub-aperture signals 1203 are fed to the unit 1208, where the sub-aperture signals are delayed for steering of receive beam direction and focusing under the assumption of a homogeneous medium with the constant, average propagation velocity, referred to as homogeneous delays. 3D beam steering and focusing can also be done with sparse arrays, where the sub-aperture unit 1202 could typically be missing. With 1.75 D arrays, the number of HF array elements can also be reduced so much that the sub-aperture units could be left out. In the following we therefore use element and sub-aperture signals synonymously.
The element signals that are corrected with the homogenous delays, 1209, are fed to a unit 1210 where corrections for the wave front aberrations are applied, for example estimated as described in Eqs. (33,34) or according to the methods described in U.S. Pat. No. 6,485,423, U.S. Pat. No. 6,905,465 and U.S. patent application Ser. No. 10/894,38, before the element signals are summed to the final receive beam signal. For 3D imaging one would use multiple receive beams with small angular offsets that covers a wide transmit beam in parallel. The aberration corrections for the angularly offset beams could be a side shifted version of the corrections for the central beam, that are added together with the homogeneous delays for the angular offset in the unit 1210.
The output 1211 of the unit 1210 is hence one or more RF-signals for one or more receive beam directions in parallel, that is fed to the processing unit 1212 according to this invention, that performs one or more of the operations according to
The aberration corrections are estimated in the unit 1213, for example according to the methods described in relation to the cited patents and patent applications and possibly also utilizing methods based on Eqs. (33,34). The unit 1213 takes as its input the homogeneously delay corrected signals 1209 and possibly also final beam signals 1214 with suppression of the pulse reverberation noise according to this invention. The delay corrected element signals 1209 are typically first processed with methods according to this invention, typically the method described in relation to
When estimation of the corrections for the wave front aberrations are based on signal correlations with the beam-former output signal 1214 with highly suppressed reverberation noise, the reverberation noise in the element signals is uncorrelated to the beam-former output signal. When slow updates of the aberration correction estimates are acceptable, one can use so long correlation times that the effect of the reverberation noise in the element signals on the correction estimates can be negligible. However, the correlation time is generally so low that it is preferable to also suppress the reverberation noise in the element signals before the estimation of the aberration corrections.
The outputs of the unit 1212 are the linearly and nonlinearly scattered signals, the two quantitative nonlinear parameters, and Doppler phase and frequency data, as described in relation to
It should be clear to any-one skilled in the art, that many simplifications of the instrument as presented in
One could also remove the estimations and the corrections for the wave front aberrations, i.e. units 1210 and 1213, and still be able to do the processing in unit 1212 to produce both linearly and nonlinearly scattered signals etc. as described above. The array could then be further simplified where elements symmetrically around the beam scan axis (the azimuth axis) is galvanically combined to further reduce the number of independent channels by a factor 2, often referred to as 1.5 D arrays. One could similarly use one dimensional (1D) arrays and also annular arrays with mechanical scanning of the beam direction, where the only modification to the block diagram in
Hence, as one can want instruments with different complexity, or selectable complexity, one also wants instruments that can select between the different methods of processing described above, for best performance according to the measurement situation. The numbering of the methods described above is related to an increase in complexity of the methods, with subsequently an increase in the number of pulses required per radial image line, and hence an increase in time per image, which is the inverse frame rate. The advantages and the drawbacks of the methods are:
The invention devices an instrument that can operate according to at least two of the methods, with the ability to select the best method for the needs, where the selection can be done under direct control of the operator, or the operator can set constraints, where the instrument automatically selects methods for best performance according to the constraints under different operating conditions.
An example constraint set by the operator can be a minimal frame rate, where for low depth ranges where it is possible to use high pulse repetition frequency, one can use the highest numbered method in the list above that still meets the frame rate constraint to obtain best possible performance with the needed frame rate. For larger depth ranges where the pulse repetition frequency must be reduced the instrument selects one of the former methods that still meets the frame rate constraint albeit with poorer estimation quality. Another example constraint is a combination of frame rate and estimation quality, where increasing the range for intermediate ranges the quality is dropped while the frame rate is maintained, and for larger depth ranges the frame rate is dropped while the quality is maintained.
The method selection could also automatically depend on imaging modality, where for linear object imaging of the heart one would use Method 1 with reverberation suppressed image signals for highest frame rate, while for studying movement in the myocardium the instrument could switch to Method 2 with 2-4 transmitted pulse complexes per radial image line, utilizing Eq. (24) for myocardial movement. For imaging of scatterer velocities the instrument could switch to Method 2 with 8-16 transmitted pulse complexes per radial image line, using the processing in unit 1215. For stationary objects like the prostate, the breast, the liver, etc. one could typically choose Method 4 for best possible estimation of the 1^{st }order linearly and nonlinearly scattered signals, the nonlinear propagation delays and quantitative image parameters.
For tomographic reconstruction, the processing according to this invention would typically be done on the individual receive element signals, before the signals are processed according to the reconstruction algorithms of various kinds, where a block schematic of a typical instrument for tomographic image reconstruction according to the invention is shown in
An example instrumentation for use of the methods for acoustic imaging of geologic structures around an oil well, is shown in
Thus, while there have shown and described and pointed out fundamental novel features of the invention as applied to preferred embodiments thereof, it will be understood that various omissions and substitutions and changes in the form and details of the devices illustrated, and in their operation, may be made by those skilled in the art without departing from the spirit of the invention.
It is also expressly intended that all combinations of those elements and/or method steps which perform substantially the same function in substantially the same ways to achieve the same results are within the scope of the invention. Moreover, it should be recognized that structures and/or elements and/or method steps shown and/or described in connection with any disclosed form or embodiment of the invention may be incorporated in any other disclosed or described or suggested form or embodiment as a general matter of design choice. It is the intention, therefore, to be limited only as indicated by the scope of the claims appended hereto.
We will in the claims use the following concepts:
Cited Patent | Filing date | Publication date | Applicant | Title |
---|---|---|---|---|
US4081783 | Sep 22, 1976 | Mar 28, 1978 | Keisuke Honda | Fish finder capable of discriminating sizes of fish |
US4483345 | Aug 4, 1982 | Nov 20, 1984 | Fujitsu Limited | Pressure measuring system with ultrasonic wave |
US4532812 | Jun 30, 1983 | Aug 6, 1985 | Nl Industries, Inc. | Parametric acoustic flow meter |
US4610255 | Nov 28, 1984 | Sep 9, 1986 | Fujitsu Limited | Ultrasonic non-linear parameter measuring system |
US4686659 | Jun 14, 1985 | Aug 11, 1987 | Keisuke Honda | Color sonar system for displaying echo signals from fish |
US4844082 | Feb 8, 1988 | Jul 4, 1989 | Matsushita Electric Industrial Co., Ltd. | Ultrasonic examination apparatus |
US4936308 | Nov 1, 1988 | Jun 26, 1990 | Agency Of Industrial Science & Technology | Method and apparatus for measuring acoustic characteristics and temperature |
US5928151 * | Aug 22, 1997 | Jul 27, 1999 | Acuson Corporation | Ultrasonic system and method for harmonic imaging in three dimensions |
US5961464 * | Sep 16, 1998 | Oct 5, 1999 | Hewlett-Packard Company | Ultrasound contrast agent detection using spectral analysis from acoustic scan lines |
US6186951 | May 26, 1999 | Feb 13, 2001 | Riverside Research Institute | Ultrasonic systems and methods for fluid perfusion and flow rate measurement |
US6226228 * | Oct 14, 1999 | May 1, 2001 | Acuson Corporation | Ultrasonic harmonic imaging system and method |
US6238342 | May 26, 1999 | May 29, 2001 | Riverside Research Institute | Ultrasonic tissue-type classification and imaging methods and apparatus |
US6312383 * | Dec 11, 2000 | Nov 6, 2001 | Riverside Research Institute | Dual band ultrasonic systems |
US6401539 | Jun 21, 2000 | Jun 11, 2002 | Acuson Corporation | Ultrasonic imaging aberration correction system and method |
US6423007 | Dec 11, 2000 | Jul 23, 2002 | Riverside Research Institute | Ultrasonic systems and methods for contrast agent concentration measurement |
US6485423 * | Jan 31, 2001 | Nov 26, 2002 | Bjorn A. J. Angelsen | Correction of phasefront aberrations and pulse reverberations in medical ultrasound imaging |
US6488626 | Apr 6, 2000 | Dec 3, 2002 | Riverside Research Institute | Ultrasonic sensing by induced tissue motion |
US6494839 * | Aug 6, 2001 | Dec 17, 2002 | Koninklijke Philips Electronics N.V. | Ultrasonic diagnostic imaging system transmitter for sum and difference frequency imaging |
US6514204 | Jul 20, 2001 | Feb 4, 2003 | Riverside Research Institute | Methods for estimating tissue strain |
US6533726 | Aug 8, 2000 | Mar 18, 2003 | Riverside Research Institute | System and method for ultrasonic harmonic imaging for therapy guidance and monitoring |
US6676599 | Jan 22, 2002 | Jan 13, 2004 | G.E. Vingmed Ultrasound As | Method and apparatus for providing real-time calculation and display of tissue deformation in ultrasound imaging |
US6682487 | Sep 10, 2002 | Jan 27, 2004 | Koninklijke Philips Electronics N.V. | Ultrasonic imaging aberration correction using harmonic and non-harmonic signals |
US6726627 | Jan 24, 2003 | Apr 27, 2004 | Riverside Research Institute | System and method for ultrasonic harmonic imaging for therapy guidance and monitoring |
US6789018 | Aug 29, 2003 | Sep 7, 2004 | Nonlinear Seismic Imaging, Inc. | Mapping reservoir rocks using frequency spectral broadening and the presence of the slow-wave |
US6846290 | May 14, 2003 | Jan 25, 2005 | Riverside Research Institute | Ultrasound method and system |
US7198601 | Feb 1, 2002 | Apr 3, 2007 | Hitachi Medical Corporation | Ultrasonic contrast medium imaging apparatus and method |
US20020032382 * | Nov 20, 2001 | Mar 14, 2002 | Michalakis Averkiou | Ultrasonic nonlinear imaging at fundamental frequencies |
US20020040188 * | Aug 6, 2001 | Apr 4, 2002 | Michalakis Averkiou | Ultrasonic diagnostic imaging of nonlinearly intermodulated and harmonic frequency components |
US20020122352 | Mar 1, 2001 | Sep 5, 2002 | Khan Tawassul A. | Mapping subsurface fractures using nonlinearity measurements |
US20020161300 | Oct 19, 2001 | Oct 31, 2002 | Lars Hoff | Ultrasound measurement techniques for bone analysis |
US20030073905 | Oct 17, 2001 | Apr 17, 2003 | Bernardi Richard Bruce | Apparatus and method for indicating mechanical stiffness properties of body tissue |
US20030114758 * | Dec 19, 2001 | Jun 19, 2003 | Jensen Seth E. | Combined fundamental and harmonic ultrasonic imaging at low MI or deeper depths |
US20040030251 * | May 9, 2003 | Feb 12, 2004 | Ebbini Emad S. | Ultrasound imaging system and method using non-linear post-beamforming filter |
US20040236222 * | May 19, 2003 | Nov 25, 2004 | Zuhua Mao | Spread spectrum coding for ultrasound contrast agent imaging |
US20040267129 | May 21, 2004 | Dec 30, 2004 | Angelsen Bjorn A.J. | Ultrasonic contrast agent detection and imaging by low frequency manipulation of high frequency scattering properties |
US20060058677 | Apr 3, 2003 | Mar 16, 2006 | Kazutaka Okada | Ultrasonograph |
EP0279314A1 | Feb 8, 1988 | Aug 24, 1988 | Matsushita Electric Industrial Co., Ltd. | Ultrasonic examination apparatus |
JPH03110491A | Title not available | |||
RU2205425C1 | Title not available | |||
WO2006009469A2 | Jul 25, 2005 | Jan 26, 2006 | Bjoern A J Angelsen | Ultrasound imaging using non-linear manipulation of forward propagation |
Reference | ||
---|---|---|
1 | Notice of Allowance dated Nov. 30, 2009 issued in corresponding U.S. Appl. No. 10/864,992. | |
2 | Office Action dated Apr. 30, 2009 issued in corresponding U.S. Appl. No. 10/864,992. | |
3 | Office Action dated Sep. 3, 2008 issued in corresponding U.S. Appl. No. 10/864,992. | |
4 | XP 000032893 Flax et al. "Phase-Aberration Correction Using Signals From Point Reflections and Diffuse Scatterers: Basic Principles" pp. 758-767, Nov. 1988. | |
5 | XP 000621081 Fukukita et al. "Ultrasound pulse reflection mode measurement of nonlinearity parameter B/A and attenuation coefficient" pp. 2775-2782, Nov. 7, 1996. | |
6 | XP 001205482 Nightingale et al. "Investigation of Real-Time Remote Palpation Imaging" pp. 113-119, Feb. 21, 2001. |
Citing Patent | Filing date | Publication date | Applicant | Title |
---|---|---|---|---|
US8939909 | Oct 29, 2012 | Jan 27, 2015 | Decision Sciences International Corporation | Spread spectrum coded waveforms in ultrasound imaging |
US8968205 * | Feb 10, 2011 | Mar 3, 2015 | Siemens Medical Solutions Usa, Inc. | Sub-aperture control in high intensity focused ultrasound |
US9022939 * | Mar 2, 2006 | May 5, 2015 | Koninklijke Philips N.V. | Microbubble generating technique for phase aberration correction |
US9081116 | Dec 11, 2012 | Jul 14, 2015 | Harris Corporation | Subterranean mapping system including spaced apart electrically conductive well pipes and related methods |
US9091776 | Dec 11, 2012 | Jul 28, 2015 | Harris Corporation | Subterranean mapping system including electrically conductive element and related methods |
US9194844 * | Jul 14, 2011 | Nov 24, 2015 | Fraunhofer-Gesellschaft Zur Foerderung Der Angewandten Forschung E.V. | Destruction-free and contactless inspection method and inspection apparatus for surfaces of components with ultrasound waves |
US9204859 | Apr 22, 2011 | Dec 8, 2015 | University Of Washington Through Its Center For Commercialization | Ultrasound based method and apparatus for stone detection and to facilitate clearance thereof |
US9239373 * | Nov 16, 2011 | Jan 19, 2016 | Siemens Medical Solutions Usa, Inc. | Adaptive image optimization in induced wave ultrasound imaging |
US20080208059 * | Mar 2, 2006 | Aug 28, 2008 | Koninklijke Philips Electronics, N.V. | Microbubble Generating Technique For Phase Aberration Correction |
US20090178483 * | Jan 9, 2009 | Jul 16, 2009 | Angelsen Bjorn A J | Nonlinear Elastic Imaging With Two-Frequency Elastic Pulse Complexes |
US20120209150 * | Aug 16, 2012 | Siemens Medical Solutions Usa, Inc. | Sub-Aperture Control in High Intensity Focused Ultrasound | |
US20130111995 * | Jul 14, 2011 | May 9, 2013 | Bernd Koehler | Destruction-free and contactless inspection method and inspection apparatus for surfaces of components with ultrasound waves |
US20130123630 * | May 16, 2013 | Siemens Medical Solutions Usa, Inc. | Adaptive Image Optimization in Induced Wave Ultrasound Imaging | |
US20130279294 * | Oct 3, 2012 | Oct 24, 2013 | Surf Technology As | Nonlinear imaging with dual band pulse complexes |
CN103346992A * | Jun 26, 2013 | Oct 9, 2013 | 中国科学院计算技术研究所 | Doppler changing rate estimating method and system in OFDM system |
WO2013051943A1 | Oct 3, 2012 | Apr 11, 2013 | Surf Technology As | Nonlinear imaging with dual band pulse complexes |
WO2014037466A1 | Sep 5, 2013 | Mar 13, 2014 | Surf Technology As | Instrument and method for ultrasound mediated drug delivery |
U.S. Classification | 600/437, 73/586 |
International Classification | G01N29/22, G01S15/89, A61B8/00, A61B8/14, A61B8/08, G01S7/52, A61B8/12 |
Cooperative Classification | G01S7/52038, G01S7/52022, A61B8/0883, A61B8/0891, A61B8/481, A61B8/14, A61B8/485, A61B8/488, A61B8/483, G01S7/52077, G01S15/8925, G01S7/52095, G01S15/8927, G01S15/8952, G01S7/52042, G01S7/52049 |
European Classification | G01S7/52S2F1, A61B8/14, G01S15/89D3A, A61B8/48B, A61B8/48F, A61B8/48D, A61B8/48J, G01S7/52S2F2, G01S15/89D1C5, G01S7/52S1B1, G01S7/52S14F |
Date | Code | Event | Description |
---|---|---|---|
Jun 29, 2011 | AS | Assignment | Owner name: SURF TECHNOLOGY AS, NORWAY Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:ANGELSEN, BJORN A.J.;HANSEN, RUNE;OYVIND, STANDAL;REEL/FRAME:026520/0727 Effective date: 20110629 |
Mar 18, 2015 | FPAY | Fee payment | Year of fee payment: 4 |